DETERMINING A SET OF SURROGATE
PARAMETERS TO EVALUATE URBAN
STORMWATER QUALITY
Nadeeka Sajeewani Miguntanna
B.Sc. (Civil Engineering, Honours)
A THESIS SUBMITTED IN FULMILMENT OF THE
REQUIREMENT FOR THE DEGREE OF MASTER OF SCIENCE IN
ENGINEERING
FACULTY OF BUILT ENVIRONMENT AND ENGINEERING
QUEENSLAND UNIVERSITY OF TECHNOLOGY
October- 2009
i
KEYWORDS
Urban water quality, urban water pollution, pollutant build-up, pollutant wash-off,
stormwater pollution mitigation, surrogate parameters.
ii
iii
ABSTRACT
This thesis details methodology to estimate urban stormwater quality based on a set
of easy to measure physico-chemical parameters. These parameters can be used as
surrogate parameters to estimate other key water quality parameters. The key
pollutants considered in this study are nitrogen compounds, phosphorus compounds
and solids. The use of surrogate parameter relationships to evaluate urban
stormwater quality will reduce the cost of monitoring and so that scientists will have
added capability to generate a large amount of data for more rigorous analysis of key
urban stormwater quality processes, namely, pollutant build-up and wash-off. This in
turn will assist in the development of more stringent stormwater quality mitigation
strategies.
The research methodology was based on a series of field investigations, laboratory
testing and data analysis. Field investigations were conducted to collect pollutant
build-up and wash-off samples from residential roads and roof surfaces. Past
research has identified that these impervious surfaces are the primary pollutant
sources to urban stormwater runoff. A specially designed vacuum system and rainfall
simulator were used in the collection of pollutant build-up and wash-off samples.
The collected samples were tested for a range of physico-chemical parameters. Data
analysis was conducted using both univariate and multivariate data analysis
techniques.
Analysis of build-up samples showed that pollutant loads accumulated on road
surfaces are higher compared to the pollutant loads on roof surfaces. Furthermore, it
was found that the fraction of solids smaller than 150 µm is the most polluted
particle size fraction in solids build-up on both roads and roof surfaces. The analysis
of wash-off data confirmed that the simulated wash-off process adopted for this
research agrees well with the general understanding of the wash-off process on urban
impervious surfaces. The observed pollutant concentrations in wash-off from road
surfaces were different to pollutant concentrations in wash-off from roof surfaces.
Therefore, firstly, the identification of surrogate parameters was undertaken
separately for roads and roof surfaces. Secondly, a common set of surrogate
iv
parameter relationships were identified for both surfaces together to evaluate urban
stormwater quality.
Surrogate parameters were identified for nitrogen, phosphorus and solids separately.
Electrical conductivity (EC), total organic carbon (TOC), dissolved organic carbon
(DOC), total suspended solids (TSS), total dissolved solids (TDS), total solids (TS)
and turbidity (TTU) were selected as the relatively easy to measure parameters.
Consequently, surrogate parameters for nitrogen and phosphorus were identified
from the set of easy to measure parameters for both road surfaces and roof surfaces.
Additionally, surrogate parameters for TSS, TDS and TS which are key indicators of
solids were obtained from EC and TTU which can be direct field measurements.
The regression relationships which were developed for surrogate parameters and key
parameter of interest were of a similar format for road and roof surfaces, namely it
was in the form of simple linear regression equations. The identified relationships for
road surfaces were DTN-TDS:DOC, TP-TS:TOC, TSS-TTU, TDS-EC and TS-
TTU:EC. The identified relationships for roof surfaces were DTN-TDS and TS-
TTU:EC. Some of the relationships developed had a higher confidence interval
whilst others had a relatively low confidence interval. The relationships obtained for
DTN-TDS, DTN-DOC, TP-TS and TS-EC for road surfaces demonstrated good near
site portability potential.
Currently, best management practices are focussed on providing treatment measures
for stormwater runoff at catchment outlets where separation of road and roof runoff
is not found. In this context, it is important to find a common set of surrogate
parameter relationships for road surfaces and roof surfaces to evaluate urban
stormwater quality. Consequently DTN-TDS, TS-EC and TS-TTU relationships
were identified as the common relationships which are capable of providing
measurements of DTN and TS irrespective of the surface type.
v
TABLE OF CONTENTS
Chapter 1 - Introduction ...........................................................................................1
1.1 Background ..................................................................................................1
1.2 Aims and objectives .....................................................................................2
1.3 Hypothesis....................................................................................................2
1.4 Justification for the research ........................................................................3
1.5 Methodology and research plan ...................................................................4
1.6 Scope............................................................................................................5
1.7 Structure of the thesis...................................................................................6
Chapter 2 - Impacts of Urbanisation........................................................................9
2.1 Background ..................................................................................................9
2.2 Hydrologic and water quality impacts of urbanisation ..............................10
2.3 Primary water pollutants in an urban environment ....................................20
2.3.1 Suspended solids ....................................................................................20
2.3.2 Organic carbon.......................................................................................22
2.3.3 Nutrients.................................................................................................23
2.3.4 Heavy metals..........................................................................................25
2.3.5 Hydrocarbons .........................................................................................28
2.4 Stormwater pollutant processes..................................................................29
2.4.1 Pollutant build-up...................................................................................30
2.4.2 Pollutant wash-off ..................................................................................35
2.5 Summary ....................................................................................................39
Chapter 3 - Mitigation Actions and Stormwater Quality Monitoring................41
3.1 Background ................................................................................................41
3.2 Current stormwater quality mitigation actions...........................................42
vi
3.3 Stormwater quality monitoring and issues.................................................46
3.4 Surrogate water quality parameters ...........................................................52
3.5 Summary....................................................................................................57
Chapter 4 - Research Tools.....................................................................................59
4.1 Background................................................................................................59
4.2 Vacuum collection system .........................................................................60
4.2.1 Selection of vacuum system...................................................................60
4.2.2 Sampling efficiency ...............................................................................62
4.3 Rainfall simulator.......................................................................................64
4.3.1 Calibration of the rainfall simulator.......................................................66
4.3.2 Calibration for rainfall intensity and uniformity of rainfall...................67
4.3.3 Drop size distribution and kinetic energy of rainfall .............................69
4.4 Model roofs................................................................................................74
4.5 Data analytical tools...................................................................................76
4.5.1 Univariate data analysis techniques .......................................................77
4.5.2 Multivariate data analysis techniques ....................................................77
4.6 Summary....................................................................................................90
Chapter 5 - Selection of Study Sites .......................................................................93
5.1 Background................................................................................................93
5.2 Study area...................................................................................................93
5.3 Study site selection ....................................................................................95
5.3.1 Investigation of road surfaces ................................................................96
5.3.2 Investigation of roof surfaces.................................................................98
5.4 Summary....................................................................................................99
vii
Chapter 6 - Sample Collection and Laboratory Testing ....................................101
6.1 Background ..............................................................................................101
6.2 Collection of samples...............................................................................102
6.2.1 Collection of pollutant build-up samples from road surfaces ..............102
6.2.2 Collection of pollutant wash-off samples from road surfaces .............103
6.2.3 Collection of pollutant build-up samples from roof surfaces...............106
6.2.4 Collection of pollutant wash-off samples from roof surfaces..............107
6.3 Treatment and transportation of water samples .......................................108
6.4 Sub sampling............................................................................................109
6.5 Laboratory testing ....................................................................................110
6.5.1 Particle size distribution.......................................................................111
6.5.2 pH, EC and turbidity ............................................................................112
6.5.3 Total suspended solids and total dissolved solids................................112
6.5.4 Total organic carbon and dissolved organic carbon.............................113
6.5.5 Nitrogen and phosphorus parameters...................................................114
6.6 Summary ..................................................................................................117
Chapter 7 - Analysis of Pollutant Build-up .........................................................119
7.1 Background ..............................................................................................119
7.2 Characteristics of build-up pollutants on road surfaces ...........................119
7.2.1 Analysis of total solids load .................................................................120
7.2.2 Particle size distribution.......................................................................121
7.2.3 Physico-chemical characteristics of build-up pollutants......................122
7.2.4 Investigation of pollutants in different particle size fractions of solids
.......124
7.3 Characteristics of build-up pollutants on roof surfaces............................126
7.3.1 Analysis of total solids load .................................................................127
7.3.2 Particle size distribution.......................................................................128
7.3.3 Physico-chemical characteristics of build-up pollutants......................129
viii
7.3.4 Investigation of pollutants in different particle size fractions of solids
.............................................................................................................131
7.4 Comparison of pollutant build-up characteristics on road surfaces and roof
surfaces ................................................................................................................134
7.5 Conclusions..............................................................................................139
Chapter 8 - Analysis of Pollutant Wash-off.........................................................141
8.1 Background..............................................................................................141
8.2 Understanding the solids wash-off process..............................................142
8.2.1 Road surfaces .......................................................................................142
8.2.2 Roof surfaces .......................................................................................146
8.2.3 Comparison of pollutants concentrations on road and roof surfaces...149
8.3 Analysis of physico-chemical parameters ...............................................153
8.3.1 Identification of potential surrogate parameters for road surfaces ......155
8.3.2 Identification of potential surrogate parameters for roof surfaces.......171
8.4 Verification of selected surrogate parameters using PLS ........................183
8.5 Conclusions..............................................................................................187
Chapter 9 - Development of Surrogate Parameter Relationships and Validation
..................................................................................................................................189
9.1 Background..............................................................................................189
9.2 Development of parameter relationships .................................................189
9.2.1 Surrogate parameter relationships for wash-off from road surfaces....191
9.2.2 Surrogate parameter relationships for wash-off from roof surfaces ....197
9.3 Portability of the relationships.................................................................200
9.4 Common surrogate parameter relationships for road and roof surfaces ..206
9.5 Conclusions..............................................................................................209
ix
Chapter 10 - Conclusions and Recommendations ..............................................211
10.1 Conclusions..............................................................................................211
10.1.1 Analysis of pollutant build-up..............................................................212
10.1.2 Analysis of pollutant wash-off .............................................................213
10.2 Recommendations for further research ....................................................216
References ...............................................................................................................219
x
xi
LIST OF TABLES Table 2.1- Sources of heavy metals in an urban environment ...................................25
Table 2.2- Influencing factors for the quality of roof runoff .....................................27
Table 2.3- Pollutant loading rates of street surfaces for different landuses ...............33
Table 3.1- Fraction of pollutants associated with different particle size ranges-
percentage by weight................................................................................55
Table 4.1- Selected control box setting for different rainfall intensities....................68
Table 4.2- List of preference functions ......................................................................87
Table 6.1- Rainfall intensities and durations simulated during the study................104
Table 6.2- Details of the test methods used for nitrogen and phosphorus compounds
..................................................................................................................................114
Table 7.1- Amount of total solids at each study site................................................120
Table 7.2- Total pollutants loads at each study site (mg/m2)...................................123
Table 7.3- Amounts of pollutants per unit weight of total solids (mg/g).................124
Table 7.4- Average total solids load (mg/m2) ..........................................................127
Table 7.5- Pollutants loads in each build-up sample (mg/m2) .................................130
Table 7.6- Amounts of pollutants per unit weight of total solids (mg/g).................131
Table 7.7- PROMETHEE 2 ranking ........................................................................136
Table 8.1- Mean concentration and standard deviation values of measures parameters
..................................................................................................................................151
Table 8.2- Correlation matrix of physico-chemical parameters obtained from
principal component analysis.................................................................157
Table 8.3- Mean concentrations of nitrogen compounds.........................................158
Table 8.4- Mean concentrations of phosphorus compounds....................................163
Table 8.5- Correlation matrix obtained from PCA ..................................................173
Table 8.6- Mean concentrations of nitrogen compounds.........................................174
Table 8.7- Mean concentrations of phosphorus compounds....................................176
Table 8.8- Potential surrogate water quality parameters for nitrogen, phosphorus and
solids ......................................................................................................182
Table 8.9- Calibration and prediction results of models ..........................................185
Table 9.1- Surrogate parameter relationships for road surfaces ..............................192
Table 9.2- Surrogate parameter relationships for roof surfaces...............................198
xii
Table 9.3- Relationship coefficients (m) and coefficient of determination for the
regression relationships..........................................................................206
Table 9.4- Common Surrogate parameter relationships for road surfaces and roof
surfaces ..................................................................................................207
Table 10.1- Surrogate parameter relationships for road surfaces ............................215
Table 10.2- Surrogate parameter relationships for roof surfaces.............................215
Table 10.3- Common Surrogate parameter relationships for road surfaces and roof
surfaces ...............................................................................................216
xiii
LIST OF FIGURES
Figure 2.1- Changes in runoff hydrograph after urbanisation....................................12
Figure 2.2- Paritcle size distribution diagram............................................................31
Figure 2.3- Hypothetical representations of surface pollutant load over time...........37
Figure 2.4- First flush for solids and COD ................................................................38
Figure 3.1- Grab (manual) sampling..........................................................................49
Figure 3.2- Automatic sampler ..................................................................................51
Figure 4.1- The water filter system of Delonghi Aqualand model ............................62
Figure 4.2a- Section of sample road surface..............................................................63
Figure 4.2b- Section of road surface at Ceil Circuit site ...........................................63
Figure 4.3- Comparison of particle size distribution of original sample and recovered
samples.....................................................................................................64
Figure 4.4- Schematic diagram of the rainfall simulator used for the study..............66
Figure 4.5- Arrangement of rainfall simulator for the intensity calibration and
uniformity testing of rainfall simulator ....................................................68
Figure 4.6- Pellets separated into each size ranges ....................................................71
Figure 4.7- Experimental setup for drop size calibration...........................................72
Figure 4.8- Calibration curve for flour pellets ...........................................................73
Figure 4.9- Model roof surfaces used in the study.....................................................76
Figure 4.10- PROMETHEE 1: partial outranking graph ...........................................88
Figure 4.11- PROMETHEE II: ranking.....................................................................89
Figure 5.1- Location of Coomera...............................................................................94
Figure 5.2- Monitoring sites of Coomera...................................................................95
Figure 5.3- Location of study sites.............................................................................96
Figure 5.4a- Study site 1- Drumbeat Street................................................................97
Figure 5.4b- Study site 2- Ceil Circuit .......................................................................97
Figure 5.5a- Deployment of tile roof surface.............................................................98
Figure 5.5b- Deployment of steel roof surface ..........................................................99
Figure 6.1- Collection of pollutant build-up samples from road surfaces ...............103
Figure 6.2- Set-up of the rainfall simulator in the study site ...................................105
Figure 6.3- Collection of samples to polyethylene containers.................................105
Figure 6.4- Collection of pollutant build-up sample................................................106
xiv
Figure 6.5-Wash-off sample collection from the roof surfac ..................................108
Figure 6.6-The Malvern Mastersizer S instrument ..................................................111
Figure 6.7- Shimadzu TOC-VCSH Total Organic Carbon Analyzer ......................113
Figure 6.8a- Seal Discrete Analyser ........................................................................115
Figure 6.8b- SmartChem 140...................................................................................115
Figure 6.9- DR 4000 Spectrophotometer.................................................................116
Figure 6.10- Block digester......................................................................................117
Figure 7.1- Variation of particle size distribution at each site.................................121
Figure 7.2a- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for Drumbeat Street ......................125
Figure 7.2b- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for Ceil Circuit ..............................125
Figure 7.3- Cumulative particle size distribution of each build-up sample.............128
Figure 7.4a- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU1 .........................................132
Figure 7.4b- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU2 .........................................132
Figure 7.4c- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU3 .........................................133
Figure 7.5- GAIA analysis for build-up samples.....................................................137
Figure 8.1a- Variation of TS concentration with rainfall duration and intensity for
Drumbeat Street ..................................................................................143
Figure 8.1b- Variation of TS concentration with rainfall duration and intensity for
Ceil Circuit..........................................................................................143
Figure 8.2a- Variation of particle size distribution with rainfall intensity for
Drumbeat Street ..................................................................................145
Figure 8.2b- Variation of particle size distribution with rainfall intensity for Ceil
Circuit .................................................................................................145
Figure 8.3- Variation of TS concentration with rainfall duration and intensity for roof
surfaces ..................................................................................................147
Figure 8.4- Variation of particle size distribution with rainfall intensity for roof
surfaces ..................................................................................................148
Figure 8.5- Biplot for all the physico-chemical parameters for both roads and roof
surfaces ..................................................................................................152
xv
Figure 8.6- PCA biplot for all the physico-chemical parameters for road surfaces.156
Figure 8.7- PCA biplot for DTN with easy to measure parameters.........................159
Figure 8.8a- Variation of DTN with DOC for Drumbeat Street ..............................160
Figure 8.8b- Variation of DTN with TDS for Drumbeat Street...............................161
Figure 8.8c- Variation of DTN with DOC for Ceil Circuit .....................................161
Figure 8.8d- Variation of DTN with TDS for Ceil Circuit ......................................162
Figure 8.9- PCA biplot for TP with easy to measure parameters ............................164
Figure 8.10a- Variation of TP with TOC for Drumbeat Street ................................165
Figure 8.10b- Variation of TP with TS for Drumbeat Street ...................................165
Figure 8.10c- Variation of TP with TOC for Ceil Circuit........................................166
Figure 8.10d- Variation of TP with TS for Ceil Circuit...........................................166
Figure 8.11- Correlation of TSS, TDS and TS with EC and TTU...........................168
Figure 8.12a- Variation of TSS with TTU for Drumbeat Street ..............................169
Figure 8.12b- Variation of TSS with TTU for Ceil Circuit .....................................169
Figure 8.13- PCA biplot for all the physico-chemical parameters for roof surfaces172
Figure 8.14- PCA biplot for DTN with easy to measure parameters.......................174
Figure 8.15- Variation of DTN with TDS for roof surfaces ....................................175
Figure 8.16- PCA biplot for TP with easy to measure parameters ..........................176
Figure 8.17a- Variation of TP with TOC.................................................................177
Figure 8.17b- Variation of TP with TTU.................................................................178
Figure 8.18- PCA biplot for TS, TTU and EC.........................................................179
Figure 8.19a- Variation of TSS with EC..................................................................180
Figure 8.19b- Variation of TDS with TTU ..............................................................180
Figure 8.19c- Variation of TS with EC....................................................................181
Figure 8.19d- Variation of TS with TTU.................................................................181
Figure 9.1a- Relationship of DTN and TDS ............................................................193
Figure 9.1b- Relationship of DTN and DOC ..........................................................193
Figure 9.1c- Relationship of TP and TS...................................................................194
Figure 9.1d- Relationship of TP and TOC...............................................................194
Figure 9.1e- Relationship of TSS and TTU .............................................................195
Figure 9.1f- Relationship of TDS and EC................................................................195
Figure 9.1g- Relationship of TS and EC..................................................................196
Figure 9.1h- Relationship of TS and TTU ...............................................................196
xvi
Figure 9.2a- Relationship of DTN and TDS............................................................199
Figure 9.2b- Relationship of TS and EC .................................................................199
Figure 9.2c- Relationship of TS and TTU ...............................................................200
Figure 9.3a- Portability of the relationship 1- DTN- TDS relationship...................202
Figure 9.3b- Portability of the relationship 2- DTN- DOC relationship..................202
Figure 9.3c- Portability of the relationship 3- TP- TS relationship .........................203
Figure 9.3d- Portability of the relationship 4- TP- TOC relationship......................203
Figure 9.3e- Portability of the relationship 6- TDS- EC relationship......................204
Figure 9.3f- Portability of the relationship 7- TS- EC relationship .........................204
Figure 9.4a- DTN- TDS relationship.......................................................................208
Figure 9.4b- TS- EC relationship.............................................................................208
Figure 9.4c- TS- TTU relationship ..........................................................................209
xvii
LIST OF APPEDICES
Appendix A- Rainfall Simulator Calibration Data...................................................253
Appendix B- Raw Data from Field Trials................................................................263
Appendix C- Build-up Analysis...............................................................................283
Appendix D- Wash-off Analysis Data Matrices ......................................................289
Appendix E- Data Matrices for Validation ..............................................................301
xviii
xix
ABBREVIATIONS
Al - Aluminium
BMPs - Best Management Practices
BOD - Biochemical oxygen demand
BPPs - Best Planning Practices
Cd - Cadmium
COD - Chemical Oxygen demand
Cr - Chromium
Cu - Copper
D50 - Median drop size
DKN - Dissolved kjeldahl nitrogen
DNO2 - Dissolved nitrite-nitrogen
DNO3 - Dissolved nitrate nitrogen
DOC - Dissolved organic carbon
DPO4 - Dissolved Phosphates
DTN - Dissolved total nitrogen
DTP - Dissolved total phosphorous
EC - Electrical conductivity
Fe - Iron
Hg - Mercury
MCDM - Multi Criteria Decision Making Methods
MLR - Multiple linear regressions
Ni - Nickel
NO2- -Nitrite-nitrogen
NO3- - Nitrate-nitrogen
PAHs - Polycyclic aromatic hydrocarbons
Pb - Lead
PCs - Principal components
PCA - Principal Component Analysis
PLS - Partial least Square Regression
PO43- - Phosphate
QA - Quality Assurance
xx
QC - Quality Control
RPD - [Ratio of (standard error of) Performance to (standard) Deviation]
SD - Standard deviation
SECV - Standard error of cross validation
SEE - Standard error of estimate
SEP - Standard error of performance
TC - Total carbon
TDS - Total dissolved solids
TKN - Total kjeldahl nitrogen
TN - Total nitrogen
TNO2 - Total nitrite-nitrogen
TNO3 - Total nitrate- nitrogen
TOC - Total organic carbon
TP - Total phosphorus
TPO4 - Total Phosphates
TS - Total solids
TSS - Total suspended solids
TPH - Total petroleum hydrocarbons
TTU - Turbidity
WSUD - Water Sensitive Urban Design
Zn - Zinc
xxi
CERTIFICATION OF THESIS
I certify that the work reported in this thesis is entirely my own effort, except where
otherwise acknowledged. I also certify that the work is original and has not been
previously submitted for any other award, except where otherwise acknowledged.
Signature of Candidate Date
xxii
xxiii
ACKNOWLEDGEMENTS
This research involved a number of research activities including field work,
laboratory testing, data analysis and thesis writing. I wish to acknowledge and thank
several people who have helped me execute these tasks during the period of this
research.
I would like to convey my sincerest thanks to my supervisors, Dr. Prasanna
Egodawatta and Prof. Ashantha Goonetilleke for their dedicated guidance,
invaluable assistance and endless encouragement throughout the accomplishment of
this research. A very special thank is also due to Dr Serge Kokot for his expert
advice and guidance during data analysis.
I am grateful to the Faculty of Built Environment and Engineering, Queensland
University of Technology (QUT), for providing me financial support during my
candidature. The support received from Coomera body Corporate in giving me
permission to conduct field activities within their premises is gratefully
acknowledged.
Members of the technical staff at QUT provided many practical inputs to this
research. I would like to convey my special thanks to Mr. Terry Beach, Mr. Wayne
Moore, Mr. Brian Pelin and Mr. Jim Hazelman for their assistance in carrying out
the field activities. I would also like to express my gratitude to Prof. Malcolm Cox,
Mr. Bill Kwiecien, Mrs. Wathsala Kumar and Mr. Shane Russell for giving me
access to the necessary laboratory facilities and providing me with the technical
support, when I was faced with difficult circumstances in testing my water samples.
My appreciation is further extended to Mrs. Diane Kolomeitz and Mr. Peter Nelson
for the support given to me for improving my writing skills in the thesis. My
appreciation is further extended to fellow researchers, particularly, Ms. Nandika
Miguntanna, Ms. Chandima Gunawardena, Mr. Isri Manganka, Mr. Parvez Mahbub,
Mr. Manjula Dewadasa, Mr. Chanaka Abesinghe, Mr. Kanchana Rathnayaka and
Mr. Rakkitha Thillakerathne for their valuable support during my research activities.
xxiv
Last but not least, I would like to express my heartfelt gratitude to my beloved
parents, relatives and friends for the encouragement, support and care I received
during the period of this research.
xxv
DEDICATION
I wish to dedicate this thesis to my beloved parents, Mr. Piyasena Miguntanna and
Mrs. Jayanthi Wijesekara for their unconditional love and to my two sisters, Nandika
and Poshitha and my best friends Rohini and Chandi for their morale support and
motivation.
xxvi
.
1
Chapter 1 - Introduction
1.1 Background
Urban stormwater runoff has been identified as one of the most important causes of
water quality deterioration in urban areas. Various urban impervious surfaces such as
streets, driveways, roofs and parking lots produce stormwater runoff even during
small rain events. A variety of pollutants which are accumulated on these surfaces
are removed by wash-off with the stormwater runoff leading to a considerable
increase in pollutant loading to receiving water bodies (Bannerman et al. 1993;
Cordery 1977; Goonetilleke et al. 2005). The degradation of receiving water quality
due to polluted urban stormwater runoff is an important issue and impacts on a
significant proportion of the urban population (Tsihrintzis and Hamid 1997). The
deterioration of receiving water quality despoils the aesthetic value of natural water
bodies (Zoppou 2001). Furthermore, the degradation of the receiving water quality
due to polluted urban stormwater runoff can have a significant impact on human and
ecosystem well-being.
Due to the severity of the problem of polluted stormwater runoff, mitigation actions
on stormwater pollution are of crucial importance. Current stormwater pollution
mitigation actions are primarily in the form of best management practices including
constructed structures such as detention and retention basins, wetlands, grass swales
and gross pollutant traps (Barrett et al. 1995; Brodie 2007; Sara et al. 2002; Scholes
et al. 2007). However, the effectiveness of such mitigation actions is limited due to
the lack of knowledge on pollutant processes, namely, pollutant build-up and wash-
off and key water quality parameters.
The lack of knowledge on key water quality parameters and pollutant processes are
mainly attributed to difficulties in planning and conducting stormwater quality
monitoring programs (Martinez 2005; US FHWA 2001). Investigation of a large
number of water quality parameters is time consuming and resource intensive
(Kayhanian et al. 2007; Thomson et al. 1997). Furthermore, dealing with a range of
2
variables in stormwater runoff monitoring programs requires sophisticated
knowledge of these variables related to the wash-off process (Martinez 2005; Milne
2002; US FHWA 2001). On the other hand, cost effective and robust methods for the
continuous measurement of pollutant concentrations are not yet fully developed
(Grayson et al. 1996). Therefore, it is important to identify a suite of easy-to-measure
surrogate parameters which can be correlated to water quality parameters of interest.
The relationships between key water quality parameters and its surrogate parameters
will provide a convenient approach to evaluate the quality of water directly, without
carrying out resource intensive laboratory experiments. However, the utility of this
approach depends on the quality of correlations between these different sets of
parameters (Grayson et al. 1996).
1.2 Aims and objectives
The primary objective of this research project was to identify a set of parameters
which can be used as surrogate parameters to evaluate urban stormwater quality and
to develop predictive models to estimate other water quality parameters based on the
measurements of surrogate parameters.
Therefore, the major aims were to; • To identify the physico-chemical parameters which are key indicators of urban
stormwater quality.
• To identify easy to measure parameters which can be act as surrogate
parameters for the key water quality parameters.
• To develop mathematical relationships among the surrogate parameters and
key water quality parameters of interest.
1.3 Hypothesis
• Surrogate parameter relationships provide a convenient approach to evaluate
urban storm water quality directly from the field measurements rather than
costly laboratory testing.
3
• Electrical conductivity (EC), Turbidity (TTU), Total suspended solids (TSS),
Total dissolved solids (TDS), Total organic carbon (TOC), Dissolved organic
carbon (DOC) are relatively easy to measure physico-chemical parameters
which can act as surrogate parameters for nutrient parameters namely, Nitrite
nitrogen (NO2-), Nitrate nitrogen (NO3
-), Total kjeldahl nitrogen (TKN), Total
nitrogen (TN), Phosphate (PO43-) and Total phosphorus (TP).
1.4 Justification for the research
It is well understood that stormwater runoff from urban areas convey a variety of
pollutants including solids, organic matter, nutrients and heavy metals which degrade
the quality of the receiving water bodies. Consequently, greater emphasis is now
being placed on the management of stormwater quality in order to safeguard the
receiving water environment. In this context, many regulatory authorities strive to
implement stormwater management strategies. One of the common examples for
such management strategies is best management practices which includes structural
best management practices such as wetlands, swales and detention basins.
However, an essential need for these management practices is accurate knowledge of
runoff quality. Quality of stormwater is typically measured in terms of a range of
quality parameters. This knowledge is required to understand the effects of runoff on
the receiving water quality and to develop appropriate mitigation actions (Barrett et
al. 1998; Han et al. 2006). However, limited knowledge in relation to pollutant build-
up and wash-off processes and key water quality parameters severely impede the
effectiveness of the mitigation actions.
This is primarily due to the difficulties which arise in planning and conducting
stormwater quality monitoring programmes (US FHWA 2001). In general, testing
for a range of quality parameters is expensive and time consuming. Furthermore,
testing of some of these parameters gives rise to further difficulties as the test
methods may need special equipment, expert knowledge and techniques.
4
In this context, it is feasible to identify a set of easy to measure surrogate parameters
and their relationships with the other water quality parameters. These relationships
will provide a convenient approach to evaluate the quality of water directly from
field-based measurements, without having to carry out resource intensive laboratory
experiments.
1.5 Methodology and research plan
The objectives of the research project were achieved through the following steps.
The proposed research methodology is based on a series of field investigations and
laboratory testing. The research project consisted of four steps as follows.
1. Literature review;
2. Study site selection;
3. Collection and testing of water samples; and
4. Data analysis.
1. Literature review
A comprehensive literature review was carried out focussing on stormwater pollution
aspects in urban landuses and key water quality parameters, which are used in the
evaluation of water quality. The literature review primarily focused on the following
main areas:
• Urbanisation and its impacts in terms of hydrologic and stormwater quality;
• Primary urban stormwater pollutants and pollutant sources;
• Current state of knowledge in relation to pollutant build-up and wash-off
processes;
• Key indicators of urban stormwater quality;
• Stormwater quality mitigation actions;
• Stormwater quality monitoring; and
5
• Past research on easy to measure surrogate water quality parameters and their
relationships with other water quality parameters.
2. Study site selection
The study sites were selected within a catchment in Gold Coast where extensive
monitoring programs are currently in place. Therefore, outcomes generated by this
research could directly contribute to those monitoring programs. The selected study
sites would represent two road surfaces and two roof surfaces as these surfaces have
been recognised as major contributors of pollutants to urban stormwater runoff.
3. Sample collection and testing of water samples
Both build-up and wash-off samples were collected from each study site using a
specially designed vacuum system and a rainfall simulator. For the wash-off sample
collection, six rainfall intensities were chosen to be simulated at each study site.
Based on the knowledge developed from the literature review, build-up and wash-off
samples were tested for a range of physico-chemical water quality parameters.
4. Data analysis
Data analysis was carried out using both univariate and multivariate data analysis
techniques. Firstly, data analysis was carried out to identify a set of surrogate
parameters for other key water quality parameters of interest. Then mathematical
relationships were derived among the key water quality parameters and the selected
surrogate parameters. Finally, the derived relationships were validated using a
separate data set obtained from a research study currently being undertaken at QUT.
1.6 Scope
• The research was confined to residential landuses in Gold Coast area.
However, it was considered that chemical processes are independent of
regional and climatic factors and traffic conditions. Therefore, the generic
6
knowledge created is applicable outside of the landuse investigated where
different regional, climatic and traffic characteristics prevail.
• The research is confined to road surfaces and roof surfaces. It was considered
that these two surfaces represent the highest fraction of impervious surfaces
and are the major contributors of pollutants to urban stormwater runoff.
• The research was confined to pollutant build-up and wash-off samples
collected in the selected study sites using a specially designed vacuum system
and a rainfall simulator.
• The seasonal variability and the influence of traffic characteristics such as
traffic volume were not considered in both pollutant build-up and wash-off
processes.
• The research focused only on common physico-chemical water quality
parameters. Microbiological parameters were not taken into consideration.
• The developed mathematical relationships were validated only for near site
portability.
1.7 Structure of the thesis
The thesis consists of eleven Chapters. The Chapter 1 is an introduction to the
research and contains the aims and objectives of the research. The Chapter 2
provides a review of published research literature related to the research project
undertaken. Chapter 3 also provides a review of published research literature
focussing on stormwater quality monitoring and surrogate water quality parameters.
Chapter 4 outlines details of research tools used including field investigation
apparatus and analytical methods used in the research. The selection of study sites is
discussed in Chapter 5. The methodology used for sample collection and laboratory
testing of samples are discussed in Chapter 6. Chapter 7, Chapter 8 and Chapter 9 are
data analysis Chapters. Chapter 7 focuses on pollutant build-up analysis and Chapter
8 provides analysis of pollutant wash-off. The main focus of these two chapters was
to identify a set of surrogate water quality parameters for other water quality
parameters. Chapter 9 presents the surrogate parameter relationships developed in
this research and validation of those relationships. Conclusions and
recommendations from the research study are presented in Chapter 10. Chapter 11
7
provides a list of references used throughout the thesis. Appendix A-E contains
information additional to the main text in Chapters 4-9. Due references have been
provided throughout the text where appendices are provided.
8
9
Chapter 2 - Impacts of Urbanisation
2.1 Background
Urbanisation is a growing concern as the number of people living in urban areas are
increasing rapidly. “Urbanisation” itself is multidimensional and has been defined in
many different ways. It may constitute industrial, commercial, or residential
development. Urbanisation is a process which may proceed gradually. In other
words, urbanisation describes a high density of people and buildings in an area
where the amount of traffic and waste is high compared to rural areas. While
urbanisation is often an integral part of development, growth of infrastructure may
result in a wide impact on natural resources and the environment (Karn and Harada
2001; Shuster et al. 2005; Yufen et al. 2008).
With urbanisation, the number of impervious surfaces within a catchment increase
dramatically. Impervious surfaces are mainly constructed surfaces such as rooftops,
sidewalks, roads and parking lots which are covered by impermeable materials such
as asphalt, concrete and stone. These materials effectively seal surfaces, repel water
and prevent percolation. The surfaces covered by such materials are hydrologically
active because the infiltration capacity of these surfaces is low (Barnes et al. 2001).
Therefore, the impervious surfaces can produce high runoff even during minor rain
events (Shuster et al. 2005).
The increase in the volume and rate of stormwater runoff causes flooding, property
damage and erosion (Tsihrintzis and Hamid 1997). On the other hand, pollutants
which are generated from a variety of urban activities, such as transportation
activities are accumulated on impervious surfaces. With storm events these
pollutants are washed off into receiving water bodies thereby contributing high
pollutants loads. The pollutant load associated with storm runoff can be significantly
higher than that from secondary treated domestic sewage effluent (Brodie 2007;
Novotny et al. 1985). Therefore, the increase of impervious surfaces causes
10
significant changes to both the quality and quantity of stormwater runoff (Brabec et
al. 2002).
This chapter focuses on identifying the impacts of urbanisation on the water
environment, main pollutant sources and primary water pollutants. Further, this
chapter presents detailed descriptions on the two main processes, which are
important in accumulation and removal of pollutants from urban impervious
surfaces, namely, pollutant build-up and wash-off.
2.2 Hydrologic and water quality impacts of urbanisation
The impacts of urbanisation on the water environment can be discussed under two
categories:
A) Hydrologic impacts; and
B) Water quality impacts.
A) Hydrologic impacts Physical changes to catchment surfaces such as the increase in impervious surfaces
can cause major changes in catchment hydrology. According to Shuster et al. (2005),
effective impervious area is defined as all impervious surface area that is
hydraulically connected (i.e. piped) to a drainage system so as to enhance
conveyance of water away from a source area, such as city streets, or a residential
neighbourhood. Some examples of effective impervious area would include streets
with kerbs or gutters that are directly connected to an outfall. It could also be a
parking lot that produces runoff which is routed to other conveyance systems.
Furthermore, roof surfaces can contribute a high percentage of effective impervious
surfaces (Chang and Crowley 1993). Effective impervious area has a pronounced
effect on catchment hydrology. This is primarily due to the bypass of potential
storages on the landscape and the conveyance to surface waters (Lee and Heaney
2003).
11
According to Brabec et al. (2002) impervious areas which are constructed of
different materials such as asphalt and concrete make the surfaces “desertlike” in
terms of hydrology and climate. Stormwater washes over paved urban surfaces in
much the same manner as it does over a desert landscape. Intense storms over urban
and desert areas can quickly generate large volumes of runoff, even flash floods,
followed by relatively dry conditions a short time later (Christopherson 2001).
The hydrologic impacts of urbanisation are apparent in the long term. In this context,
changes to the natural water balance are significant. Waananen (1969) suggested that
due to the urbanisation, the volume of water originating from catchments in the long
term is increased. The main reason for this is a reduction in infiltration of water into
the soil and the increase in the runoff volume due to the high fraction of impervious
surfaces. Moreover, Waananen (1969) suggested that severe floods and droughts are
common to urban catchments. He noted that urban creeks which were previously
perennial can become ephemeral for significant periods of the year mainly due to the
lack of ground water recharge and consequent reduction of base flow.
According to several research findings (for example, Farahmand et al. 2006, Shuster
et al. 2005), the main hydrologic changes to urban catchments commonly are:
• Increased runoff volume;
• Increased runoff peak flow;
• Reduced time of concentration; and
• Reduced base flow.
Combined effects of these changes can be termed as changes to the natural runoff
hydrograph as shown in Figure 2.1.
12
Figure 2.1- Changes in runoff hydrograph after urbanisation
(Adapted from Chow et al. 1988)
Increase in runoff volume
Reduced infiltration due to impervious surfaces leads to greater volumes of
stormwater runoff and more rapid peak stream discharges. On a pervious surface, a
portion of rainfall infiltrates into the soil and the remainder is converted to surface
runoff. This surface runoff and perhaps a portion of infiltrated water eventually flow
into receiving water bodies. On the other hand, high proportion of impervious areas
greatly reduces the amount of water infiltrating into the soil. Therefore, a higher
proportion of the rainfall becomes surface runoff which leads to the increase in
runoff volume (Zoppou 2001). As Seaburn (1969) has noted, the volume of
stormwater runoff from urban catchments can increase from 1.1 to 4.6 times greater
than the corresponding runoff during the rural period.
Increase in runoff peak flow
Peak discharges from an urbanised area are higher than those for a rural area.
According to Harned (1988), highway areas with more than 50% impervious surface
13
exhibit a peak discharge which is increased by a factor of four and time of
concentration which is decreased by more than half compared to equivalently sized,
undeveloped areas. As the amount of impervious surface expands, a greater
proportion of the rainfall tends to appear in the drainage system as surface runoff.
Since a relatively larger runoff volume is discharged within a shorter time interval,
peak flow rates also increase (Hall and Ellis 1985). Espey et al. (1969) showed that
there is an increment of two to four times in peak runoff discharge in an urban
catchment in comparison to a similar rural catchment. Cordery et al. (1976) found
that after analysing recorded discharges for a selected catchment before and after
urbanisation, urban flood peaks were three or four times as large as the
corresponding rural floods.
Reduced time of concentration
The introduction of impervious surfaces, compacted soils, gutters, lined channels,
and pipes increase the hydraulic conveyance efficiency of urban drainage systems.
This leads to an increase in the velocity of runoff. The increased flow velocities
reduce the time required for water to gather at the outlet of the catchment. Espey et
al. (1969) after studying several small catchments found that the time of
concentration may be reduced by a factor of one third due to urbanisation, depending
on the degree of channel improvements.
Reduced stream base flow
The reduction of infiltration leads to decreased groundwater recharge. This decreases
the base flow contribution to stream channels (Pouraghniaei 2002).
B) Water quality impacts
Urban stormwater runoff has a significant impact on the quality of receiving water
bodies. When it rains, water flows over roofs, streets, driveways, sidewalks, parking
lots and other urban land surfaces. Along the way, it picks up a variety of pollutants,
which are produced by anthropogenic activities. The impact of rainfall will dislodge
the solid particles deposited on the surface. Many pollutants adhere to these solid
14
particles and are transported along with soluble pollutants by the runoff. The
momentum associated with the runoff dislodges other pollutants which are attached
to impervious surfaces. These are transported to a water body by the flowing water
and progress through the urban catchment (Brinkmann 1985; Sartor et al. 1974;
Zoppou 2001).
The polluted stormwater runoff can endanger the water quality of receiving water
bodies, making them unhealthy for people and aquatic life. This is primarily due to
the wide variety of pollutant types and the magnitude of the pollutant load carried by
the urban stormwater runoff. Sartor et al. (1974) found that the pollutant load in
urban storm runoff is significantly higher than that in the secondary treated sewage
effluent. Furthermore, Sonzogni et al. (1980) in their study noted that there were 10
to 100 times greater suspended solids and nutrients load originating from urban areas
compared to an equivalent un-urbanised area.
According to Arnold and Gibbons (1996), there are four basic qualities of
imperviousness that make it an important indicator of water quality.
1. Although the impervious surface does not directly generate pollution, a clear
link has been made between an impervious surface and the hydrologic changes
that degrade the water quality;
2. An impervious surface is a characteristic of urbanisation and urbanisation leads
to the increase in impervious surfaces in a catchment;
3. An impervious surface prevents natural pollutant processing in the soil by
preventing percolation; and
4. Impervious surfaces convey pollutants to water bodies typically through the
direct piping of stormwater.
Researchers have identified the main sources which influence the degradation of
urban stormwater quality (Brinkmann 1985; Pitt 1979; Zoppou 2001). The sources of
pollutants in an urban area include:
• Vehicular traffic;
15
• Construction and demolition activities;
• Industrial and commercial activities;
• Corrosion;
• Urban erosion ;
• Vegetation; and
• Accidental spills.
(Brinkmann 1985; Brodie 2007; Pitt et al. 1995)
Vehicular traffic
The pollutants generated due to vehicular activities are mainly attributed to road
surfaces (Brinkmann 1985; Novotny et al. 1985). Lee and Heaney (2003) found that
around 70% of the total impervious areas are transportation related such as roads,
driveways and parking lots. Consequently, road surfaces have been identified as one
of the leading causes of degradation of water quality (Barrett et al. 1998; Ellis et al.
1987). Road surfaces have a profound impact on stormwater runoff quality as they
contribute relatively high pollutant loads in an urban area (Bannerman et al. 1993,
Sartor and Boyd 1972). Hoffman et al. (1984) found that road surface runoff can
contribute up to 80% of pollutant loadings to receiving water bodies. In addition,
road surfaces provide an efficient pathway for stormwater runoff to flow to receiving
water bodies.
Vehicular traffic contributes liquid and solid materials to urban road surfaces. Road
surface runoff is an important source of organic and inorganic-pollutants, such as
heavy metals, hydrocarbons and suspended solids (Herngren et al. 2006; Shaheen
1975; Tsihrintzis and Hamid 1997). Vehicles provide a continuous input of
pollutants to road surfaces and to runoff for the duration of rainfall events. These
pollutants originate from different activities related to vehicular traffic such as:
• Vehicle combustion exhaust;
• Leakages of vehicle lubrication oils;
• Abrasion products from vehicles such as tyre wear and brake linings;
• Pavement degradation; and
16
• Atmospheric deposition.
(Brinkmann. 1985; Han et al. 2006; Shaheen 1975)
Heavy metals are important pollutants associated with automobile activity.
Automobile tyre wear is a major source of Zn in urban runoff and is mostly
deposited on street surfaces. Tyre abrasion produces pollutants like rubber, soot and
metal oxides with Zn, Pb, Cr, Cu and Ni. Brake pad abrasion produces Ni, Cr, Cu,
Pb and Fe to the road surface runoff (Pitt et al. 2004). Gobel et al. (2006) found that
the concentrations of total suspended solids (TSS) varies from 66 mg/L to 937 mg/L
and COD concentrations rise from 2.0 mg/L to 36 mg/L for urban road surfaces in
comparison to the road surfaces in a rural catchment. They suggested that this could
be due to tyre abrasion on the road surfaces. Shaheen (1975) estimated that
approximately 0.7 g/axle.km of solids on road surfaces can be directly attributed to
vehicular activities. He found that asphalt pavement wear contributes high
concentrations of heavy metals such as Zn, Cd and Pb to stormwater runoff.
The generation of pollutants and their concentration on urban road surfaces varies
widely depending on a range of factors such as road surface condition, traffic
density, wind and road maintenance activities. Most of the automobile pollutants are
deposited on parking lots and street surfaces. However, some automobile related
pollutants are also deposited in areas adjacent to the streets. This occurs naturally by
wind and traffic-induced turbulence (Pitt et al. 2004). Brinkmann (1985) found that
the accumulation of abrasion products from tyres on urban road surfaces depends on
the volume of traffic, distribution of traffic lights, road conditions and driving habits.
They found that fuels, motor oils and lubricants are spilled on roads in high
concentrations at parking lots and near traffic lights. These materials degrade with
time and when exposed to sunlight produce hydrocarbons such as Polycyclic
aromatic hydrocarbons (PAHs). Due to the importance of vehicle-related pollutants,
a number of previous studies have hypothesised that the accumulation of pollutants
is directly related to the type of traffic on the road (Ball et al. 1998; Gobel et al.
2006).
Additionally, Barrett et al. (1995) and Sartor and Boyd (1972) found that the
effectiveness of street cleaning activities and maintenance practices have a
17
significant impact on the accumulation of pollutants on urban road surfaces. Sartor
and Boyd (1972) noted that the amount of pollutants present on the road surfaces is
dependent on the time gap since the street was last cleaned either by street sweeping
or by rainfall. Furthermore, Duncan et al. (1985) suggested that removal of
pollutants from street cleaning activities varies with the mechanism which is used for
the activity. He found that vacuum sweepers remove about 70% of the pollutants
while broom sweepers remove only about 20% of pollutants which are accumulated
on road surfaces.
Road surface condition is another important factor which affects the composition of
pollutants on road surfaces. Sartor and Boyd (1972) found that asphalt pavements
contribute about 80% more pollutant loading than concrete surfaces. Furthermore,
Egodawatta et al. (2006) noted that composition of pollutants on road surfaces vary
with the surface texture after investigating three road surfaces in Gold Coast,
Queensland State, Australia.
Construction and demolition activities
Construction and demolition activities have a significant impact on urban stormwater
quality (Brinkmann 1985). Construction activities in urban areas contribute
considerable quantities of solids and litter to the urban environment. Pollutants from
construction sites are primarily in the form of dust particles or erodible solids, which
originate from brick debris or cement particles (Brinkmann 1985). According to the
US EPA (1993), solids runoff rates from construction sites are typically 10 to 20
times greater than for agricultural lands and 1000 to 2000 times greater than for
forested lands. The pollutant loading rates can vary considerably with the amount of
construction and maintenance activities and management of its sites.
Industrial and commercial activities
Industrial and commercial activities associated with urbanisation have significant
impacts on urban stormwater pollution. Several researchers (for example, Bian and
Zhu 2008) have noted that industrial areas contribute high pollutant load to urban
stormwater runoff in comparison to the other landuses. Bannerman et al. (1993)
18
found that road surfaces and parking lots are critical source areas for the generation
of pollutants in industrial and commercial areas. Furthermore, they noted that road
surfaces in industrial areas contain higher pollutant loads than roads in commercial
or residential areas. This may be probably due to the smaller incidence of street
cleaning activities in industrial areas compared to commercial and residential areas.
The pollutant generation of industrial processes mainly depends on the nature of the
industry and their management practices. The pollutants resulting from various
industrial and commercial activities can contain many chemical toxins, heavy metals,
hydrocarbons and gases. These pollutants can be generated while loading and
unloading of equipment and through spillages and leakages of industrial materials.
Pollutants generated in commercial areas are mainly attributed to gas filling stations,
parking lots and shopping centres. However, the quality and quantity of pollutant
discharge to urban stormwater runoff from industrial and commercial activities also
depends on factors such as traffic volume and degree of surface imperviousness
(Mark 1996).
Corrosion
Corrosion of structures such as gutters, roofs and fences in the urban environment
result from exposure to atmosphere, rainfall, wind abrasion or acid rain. The rate of
corrosion depends on the availability of corrodible materials, the frequency and
intensity of exposure to an aggressive environment and the structure of the material
and its maintenance. For example, corrosion rate of roof surfaces near the sea is
significantly high due to the salty nature of the atmosphere (Pringle 1998). In urban
landuses where metallic roofs are common, corrosion is a considerable source of
pollution (Brinkmann 1985). Pitt et al. (1995) found that there is a significantly
higher concentration of heavy metals in roof runoff due to the corrosion of metals on
roofs. Different metals are used for roof surfaces such as Cu, Al, Pb and Zn for roof
covering, gutters and down pipes. All these materials release heavy metals as
corrosion products. The corrosion processes are enhanced because of the low pH
value of the rainwater. Finally, the corroded particles which accumulate on the
ground and on roof surfaces are eventually washed off with rainfall and added to the
stormwater runoff (Gobel et al. 2006).
19
Erosion
Erosion from construction sites contributes large quantities of solids to urban runoff.
In most construction sites, protective vegetative cover is removed and unprotected
soil is left exposed to rainfall. This contributes to increasing the solids loading in
stormwater runoff. Hydrologic changes associated with urbanisation such as higher
peak flows have a significant impact on erosion. Furthermore, factors such as soil
type, topography, vegetation, climatic conditions and catchment management
practices significantly affect the soil erosion. This leads to an increase in the annual
solids loads originating from urban catchments (Nielson and Booth 2002).
Vegetation
Vegetative matter commonly found in urban areas includes plant materials such as
pollen, bark, twigs and grass. The input of vegetation is dependent on the catchment
characteristics and seasonal changes. Novotny et al. (1985) studied the vegetation
input in a catchment in the United States during the fall season and found that a
mature tree can produce 15 to 25 kg of leaf residue with significant amounts of
nutrients. Furthermore, they suggested that the rainfall, which penetrates the tree
canopy, is enriched with nutrients and organics. However, Allison et al. (1998) have
questioned the importance of leaf litter as a nutrient source in urban stormwater.
Based on the outcomes of their study on an urban area in Melbourne, Australia, they
found that the contribution of nutrients from leaf litter was about two orders of
magnitude smaller than the total nutrient load measured.
Accidental spills
Pollutants generated from spillages are dependent on the nature of the spill.
However, quantitative analysis of these contaminants is difficult. Researchers
(Rogge et al. 1993; Sartor and Boyd 1972) have found that vehicular activities such
as lubrication leakages are the main source of spills on urban street surfaces. Spills
can degrade the quality of stormwater physically, chemically and biologically.
However, spillages due to industrial and commercial activities can be minimised by
good maintenance and management practices.
20
2.3 Primary water pollutants in an urban environment
Many types of pollutants originating from a variety of sources accumulate over
urban impervious surfaces. These pollutants are subsequently washed into water
bodies during storm events, leading to the degradation of receiving water quality
(Bian and Zhu 2008; Gnecco et al. 2005; Goonetilleke et al. 2005). Therefore,
identification of urban stormwater pollutants and their chemical characteristics is
important. Primary water pollutants which can be found in an urban environment
are:
• Suspended solids;
• Organic carbon;
• Nutrients;
• Heavy metals; and
• Hydrocarbons.
(Bian and Zhu 2008; US EPA 1993)
2.3.1 Suspended solids
In the urban environment, loading of suspended solids to water bodies mainly occur
from the wash-off of particles. Cordery (1977) found that the loading of suspended
solids during the initial period of a storm event is significantly higher than that in
secondary treated sewage effluent. Suspended solids in urban stormwater runoff
originate from different sources. Several researchers have noted construction
activities as the largest direct source of human-derived solids loads to stormwater
runoff (Barrios 2000; Brodie 2007: Schueler 1997). Pitt (1979) found that the main
sources of suspended solids to stormwater runoff in urban areas include wet and dry
atmospheric deposition, wear of roads, vehicles and soil erosion. Furthermore,
weathering of roofing materials also contributes high amounts of suspended solids to
the stormwater runoff (Forster 1996; Forster 1999; Gadd and Kennedy. 2001; Quek
and Forster 1993). According to Forster (1999), roof surface characteristics such as
smooth surfaces and steeper slopes have low resistance and tend to contribute higher
concentrations of solids particles to the roof runoff.
21
Accumulation of suspended solids in receiving water bodies can be harmful for
aquatic species because of their ubiquitous nature. The presence of suspended solids
increases turbidity, reducing the amount of light penetration, retards photosynthesis
and hence, may lead to decreases in food supply to aquatic life. Solids particles can
also clog water treatment plant filters, block channels and pipes, causing flooding
and property damage (Atasoy et al. 2006).
Chemical impact of suspended solids is a significant issue. Many researchers have
noted that pollutants such as hydrocarbons, heavy metals and nutrients are bound to
the suspended solids (for example, Atasoy et al. 2006; Bian and Zhu 2008; Ongley et
al. 1981; Viklander 1998). Dong et al. (1984) noted that metals associated with the
coarse fractions of suspended solids settle more quickly while those associated with
the finer fractions stay in suspension longer. Therefore, the pollutants attached to the
finer fraction have a greater impact on water quality.
It is important to have a clear understanding of the amount of pollutants associated
with different particle size ranges so that treatment facilities can be effectively
designed to target the most polluted particle sizes. Vaze and Chiew (2002) found that
although more than half of the solids are coarser than 300 µm, <15% of the total TP
and TN are attached to particle sizes >300 µm. This finding suggested that, to
effectively reduce nutrient loads in particulates, treatment facilities must be able to
remove the finer particles (down to 50 µm for TP and down to 10 µm for TN) and
not just the total suspended solids load (Vaze and Chiew 2002). Moreover, Ellis et
al. (1981) found that the greatest mass of suspended solids in urban runoff typically
occurs in the 1-50 µm particle size range. On the other hand, Herngren et al. (2005)
showed that the majority of suspended solids transported in urban runoff are below
150 µm. Furthermore, they suggested that majority of the pollutants are associated
with this size range. Due to this reason Herngren et al. (2005) suggested that street
cleaning programs should focus on removing particles below 150 µm.
The loading of suspended solids in urban stormwater runoff varies with several
factors such as storm duration and rainfall intensity. Deletic (1998) observed that the
concentration of suspended solids decreases with storm duration, only during long
and very intense rainfall for the catchment they studied. They suggested that enough
22
solids are usually available on the surface to be picked up, except during large
storms. Additionally, Egodawatta et al. (2006), after analysing water quality and
runoff data for a mixed urban catchment in the Gold Coast, Queensland, Australia
noted that average rainfall intensity strongly correlates with the total suspended
solids load. Additionally, according to Williamson (1986), suspended solids
concentrations and hence loads in the stormwater runoff at the catchment outlet may
vary a great deal between catchments depending on the potential carrying capacity of
the drainage systems and the availability of transportable material.
2.3.2 Organic carbon
Organic carbon is an oxygen demanding material. Organic carbon in the form of
grass clippings, leaves, animal waste and street litter is commonly found in urban
stormwater runoff. The major impact of organic carbon is due to its decomposition
process as it consumes dissolved oxygen in the water which is essential to fish and
other aquatic life for their survival. In addition, oxygen depletion can affect the
release of toxic chemicals and nutrients from solids deposited in a water body
(Zoppou 2001).
Researchers such as Gromaire-Mertz et al. (1999) and Sartor and Boyd (1972) found
that large amounts of organic carbon materials are added to stormwater runoff from
street surfaces. They noted that the accumulation of these materials on urban
surfaces is dependent on the time elapsed since the last street cleaning, or rainfall
and landuse characteristics. They also noted that accumulation of organic material on
street surfaces is much faster in comparison to the accumulation of inorganic
materials.
According to Sartor and Boyd (1972), finer particles of suspended solids contain
more organic carbon than the coarser particles. They also suggested that suspended
solids which contain organic carbon can easily break down into fine particulates
because of the low structural strength. Roger et al. (1998) noted that the organic
carbon concentration is high in particles smaller than 50 µm, in comparison to the
other particle sizes in road surface runoff. Moreover, organic carbon adsorbed on
23
suspended solids increases their adsorption capacity for combining with other
materials such as heavy metals (Parks and Baker 1997).
Researchers have noted that organic carbon is a good indicator of urban stormwater
quality (Han et al. 2006; Zoppou 2001). According to Zoppou (2001), there are three
common measurements of oxygen demand. These are chemical oxygen demand
(COD), biochemical oxygen demand (BOD) and total organic carbon (TOC). TOC is
a more convenient and direct expression of total organic content in a sample. TOC
can be used to estimate the accompanying BOD or COD in a sample if a relationship
can be established between TOC and BOD or COD. TOC is independent on the
oxidation state of the organic matter and does not measure other organically bound
elements such as nitrogen, hydrogen and inorganic substances which can contribute
to the oxygen demand measured by BOD and COD (APHA 2005).
2.3.3 Nutrients
Runoff from urban areas contains nutrients such as nitrogen, phosphorus, carbon,
calcium, potassium, iron and manganese. These nutrients originated from different
sources such as fertilizer applications, plant matter, vehicular activities and
atmospheric deposition (Brezonik and Stadelmann 2002; US EPA 1999). Pitt et al.
(2004) found that lawns could contribute more than 50% of the annual total
phosphorus load in a residential area. Power and Schepers (1989) and Makepeace et
al. (1995) suggested that nitrogen and phosphorus compounds in urban stormwater
runoff originate from vehicle exhausts. The potential contribution of atmospheric
deposition to nutrient loading on urban catchment surfaces is also important.
Atmospheric deposition typically supplies as much nitrogen as is washed off in
urban runoff, and smaller proportions of suspended solids, phosphorus and heavy
metals (Walsh 2000). According to a study carried out by the Nationwide Urban
Runoff Program of USA (NURP), atmospheric deposition accounts for
approximately 70% - 95% of the nitrogen and 20% - 35% of the phosphorous in
urban runoff (Schueler et al. 1991). Line et al. (2002) after studying the data
gathered from several studies, found that the annual average export values from
24
urban areas range from 1.6 to 38.5 kg/ha for nitrogen and from 0.03 to 6.23 kg/ha for
phosphorus.
Increased levels of nutrients create major problems in receiving water bodies. Excess
of nitrogen and phosphorus can stimulate aquatic life to the extent that plant growth
becomes a major problem. Excessive plant growth can choke water bodies and lead
to large fluctuations in dissolved oxygen levels (US EPA 1999). Also excessive
growth of these organisms can clog water intakes and block sunlight to deeper
waters. This seriously affects the respiration of aquatic invertebrates, leading to a
decrease in animal and plant diversity and affects the use of the water for drinking,
fishing, swimming and boating (Pouraghniaei 2002). Nitrogen and phosphorus have
significant impacts on degradation of water quality when compared to nutrients such
as calcium and potassium (Atasoy et al. 2006; Vaze and Chiew 2002).
Nitrogen can be available in stormwater runoff as inorganic and organic nitrogen.
The most important forms of inorganic nitrogen in terms of their immediate impact
on water quality are the readily available ammonia ions (NH4+ and NH3), nitrites
(NO2-) and nitrates (NO3
-) (US EPA 1999). Total kjeldahl nitrogen comprises of
organic form of nitrogen. Total Nitrogen (TN) is the sum of all these forms of
nitrogen. Furthermore, nitrogen can be converted between these forms and also to
nitrogen gas, by chemical and biological action. Nitrogen can be transported in
surface runoff in both particulate and dissolved phases (Lee and Bang 2000; Taylor
et al. 2005).
Phosphorus in stormwater can exist in organic or inorganic forms, and also
phosphorus can be available in particulate or dissolved phases. Total phosphorus is
the sum of dissolved and particulate phosphorus. Each of these fractions can be
subdivided into reactive, acid-hydrolysable and organically bound phosphorus,
according to its chemical availability. Reactive phosphorus is readily available, while
organic phosphorus is released only by powerful oxidising agents. As phosphorus
has an ability to adsorb to soil particles and organic matter, it is transported in
surface runoff with eroded solids. The sorbed phosphorus can enter runoff by both
dry fallout and rainfall (US EPA 1999).
25
2.3.4 Heavy metals
Urban stormwater runoff contains significant amounts of heavy metals. Tsihrintzis
and Hamid (1997) found that the concentration of heavy metals in stormwater runoff
is one or two orders of magnitude greater than that in sanitary sewage. In comparison
to the other pollutants, heavy metals do not degrade in the environment. Therefore,
releasing of metals into receiving water bodies has a significant impact on water
quality.
Commonly found heavy metals in urban stormwater runoff are Copper (Cu), lead
(Pb), zinc (Zn), mercury (Hg) and cadmium (Cd). The primary sources of these
metals in stormwater runoff are discussed in Table 2.1 below.
Table 2.1- Sources of heavy metals in an urban environment (Adapted from Charlesworth et al. 1999)
Metal Sources
Cadmium- (Cd)
batteries, pigments and paints, plastics, printing and graphics, wear of car tyres corrosion of metals, fossil fuel combustion, medical uses, metallurgical industries
Nickel - (Ni) batteries, metallurgical industries
Zinc- (Zn)
wear of car tyres, corrosion of metals, fossil fuel combustion, electronics batteries, pigments and paints, plastics, printing graphics, medical uses metallurgical industries
Copper- (Cu) electronics waste, metallurgical industries
Lead- (Pb)
fossil fuel combustion, batteries, pigments and paints, printing graphics, medical uses, metallurgical industries
26
Industrial and commercial landuses are the greatest contributors of heavy metals to
runoff. A study carried out by Brezonik and Stadelmann (2002) found that
commercial and industrial landuses in Minnesota, USA contributed a higher amount
of heavy metals than a residential landuse. Sartor and Boyd (1972) noted that metal
concentrations in street sweepings are considerably higher for industrial areas.
Moreover, from a study carried out in three different landuses in Queensland,
Australia, Herngern et al. (2006) noted that industrial sites have the highest
concentration of heavy metals, in comparison to the residential and commercial
landuses.
Vehicular and traffic related activities are another important source of heavy metals
to urban stormwater runoff. A study carried out by Sansalone et al. (1996) suggested
that tyre wear, brake wear and fuel leakage are common sources of heavy metals
which are generated from traffic related activities. Pb concentrations on urban road
surfaces can increase due to the particles of paint from road markings. Furthermore,
Sansalone et al. (1996) suggested that the concentrations of these metals such as Zn,
Cu and Cd are different in the middle of the lane and near the kerb due to vehicular
movement. However, according to Gobel et al. (2006) different traffic densities have
different effects on pollutant distribution and concentration.
Roof surfaces are also identified as a significant source of heavy metals to urban
stormwater runoff (Gobel et al. 2006). Roof surfaces account for about half of the
total runoff volume from impermeable surfaces in urban areas of industrialised
countries. Metal roofs, such as Cu and Zn have the highest impact on heavy metal
concentration in roof runoff (Mosley and Peake 2001). Gobel et al. (2006) found that
heavy metal concentrations in stormwater originating from roofs with Zn gutters and
downpipes and metal roofs made out of Cu and Zn have higher heavy metal
concentrations in stormwater than urban street surfaces with heavy traffic. This is
mainly due to the corrosion of metallic components on roof surfaces. The corrosion
process is enhanced because of the low pH of rainwater. According to Mosley and
Peake (2001), influencing factors for the quality of roof runoff can be summarised as
follows:
27
Table 2.2- Influencing factors for the quality of roof runoff (Adapted from Mosley and Peake 2001)
Parameter Factors
Roof material chemical characteristics, roughness,
surface coating, age, weathering
Physical boundary conditions of the
roof
size, inclination, exposure
Precipitation event
intensity, wind, pollutant concentrations
in the rain
Chemical properties of the
substance
vapour pressure, partition coefficient,
solubility in water
Other meteorological factors
season, air masses, duration and weather
characteristics of antecedent dry period
These metals in stormwater runoff can exist as solids or liquid. Brinkmann (1985)
found that an appreciable amount of heavy metals in urban stormwater runoff is
transported in the solute phase and chemically, they may be organic or inorganic
compounds. Some heavy metals such as Cu, Cd, Pb and Zn are more soluble in water
than others and may cause toxic effects. Connel (1993) noted that the presence of
other metals, temperature and salinity can enhance the toxicity of a specific metal.
Baird (1999a) suggested that physico-chemical parameters such as pH and the
amount of dissolved and organic carbon present in the water can lead to interactions
such as adsorption of metal ions in stormwater runoff.
Researchers have noted that heavy metals in urban stormwater runoff are strongly
attached to suspended solids (Dong et al.1984; Ujevic et al. 2000). As Deletic and
Orr (2005) noted, there is always a difference between the concentrations of each
metal measured in different size fractions of suspended solids. Ujevic et al. (2000)
found that the concentrations of heavy metals such as Cr and Pb in particle size
fractions < 54 µm were four times greater than in the coarse fraction. According to
them, on average, the 2-63 µm fraction contained eight to ten times greater
28
concentrations of Zn, Cu and Cd than the >500 µm fraction. Moreover, Roger et al.
(1998) showed that Zn was most often found in particles <50 µm, while Pb
concentrations were similar in all particle size fractions that they analysed. In
contrast, Wang et al. (1981) found that in stormwater, 92% of the Pb was associated
with particles larger than 20 µm in diameter. However, Dong et al. (1984) analysing
chemical characteristics of different particle size fractions of suspendered solids
noted that all metals are not concentrated in the fine fraction of solids. They noted
that, Cr, Cu, Fe and Ni in urban street dust samples were more evenly distributed
among all particle size fractions suggesting that they were derived from abrasion of
metal surfaces with varying degrees of corrosion (Dong et al. 1984).
2.3.5 Hydrocarbons
Urban stormwater runoff contains a wide array of hydrocarbon compounds (Ball et
al. 2000). Relatively high concentrations of hydrocarbons are generated from roads,
parking lots, vehicle service stations and residential parking areas (Connecticut
2004). Latimer et al. (2004) found that most hydrocarbons are generated because of
leakages of crankcase oil associated with vehicular traffic with street dust, soil and
atmospheric deposition being minor sources of hydrocarbons. In the study by
Latimer et al. (2004), hydrocarbons originating from sources such as street dust, soil
and atmospheric deposition along the highway were analysed. It was concluded that
these sources could be responsible for only 12% of the hydrocarbons in highway
runoff with the remaining 88% of the hydrocarbons originating from crankcase oil
drips in the centre of each travel lane. Hunter et al. (1979) estimated that 4.2×109 L
of automobile and industrial lubricants are lost to the environment annually in the
USA either by direct disposal to sewers and application to land or indirectly by
spillage or leakage from motor vehicles (Hunter et al.1979).
Polycyclic aromatic hydrocarbons (PAHs) are one of the main types of hydrocarbons
that influence the deterioration of water quality (Van Metre et al. 2000). PAHs are
mainly generated from incomplete combustion of fossil fuels. The chemical structure
of PAHs is a combination of benzene rings with linear or branched arrangement.
many studies have indicated that PAHs have low solubility in water due to their
29
stable chemical structure (for example, Marsalek et al. 1997). Hoffman et al. (1984)
suggested that the generation of PAHs in urban stormwater runoff is dependent on
factors such as landuse and antecedent dry period. Gobel et al. (2006) suggested that
PAH concentrations in stormwater runoff increase with the rainfall intensity.
Van Metre et al. (2000) have noted that the use of automobiles significantly
increases the concentration of PAHs in stormwater runoff. Ellis (1986) noted that
70% of the total PAHs found in receiving waters can be attributed to highway runoff
sources. Steuer, et al. (1997) found that PAH levels in commercial parking lots were
10 to 100 times higher than that from any other source area. They concluded that
even though the commercial parking lots represent only 3% of an urban drainage
area, it can contribute around 60% of the annual PAH load to urban stormwater
runoff. In addition, hydrocarbons are generated from natural sources, such as forest
fires and degradation of organic materials (Barbara et al. 2009; Chernova et al.
2001).
Increased concentration of hydrocarbons leads to an increase in the toxicity in
receiving water bodies. Therefore, excess amounts of hydrocarbons can be harmful
for aquatic life. In addition, large quantities of hydrocarbons can affect drinking
water supplies and the recreational use of water (Connecticut 2004).
Hydrocarbons in urban stormwater runoff are associated with particulate matter.
Vaze and Chiew (2002) noted that petroleum hydrocarbons are strongly associated
with suspended solids in highway runoff. They concluded that 88% to 96% of the
total hydrocarbons discharged into the stormwater runoff are particulate
hydrocarbons. According to Datry et al. (2003) hydrocarbons in dissolved form are
rarely found.
2.4 Stormwater pollutant processes
Accumulation and removal of pollutants from urban impervious surfaces is a very
complex process. This process is described and modelled using two main concepts.
30
These are;
• Pollutant build-up
The concept of defining the processes related to accumulation of pollutants on
impervious surfaces.
• Pollutant wash-off
The concept of defining the processes involved in removal of accumulated
pollutants from catchment surfaces by rainfall and runoff.
(Duncan 1995; Egodawatta and Goonetilleke 2006; Vaze and Chiew 2002)
2.4.1 Pollutant build-up
Build-up is the process by which dry deposition accumulates on impervious surfaces.
Build-up on impervious surfaces can be described as a dynamic process, where there
is equilibrium between deposition and removal and between pollutant sources and
sinks at any given time (Duncan 1995). The build-up of pollutants mainly depends
on factors such as,
• Antecedent dry days;
• Wind speed;
• Landuse;
• Traffic;
• Population density;
• Street cleaning practices; and
• Pavement material and condition.
(Egodawatta 2007; Pitt 1979; Sartor and Boyd 1972; Zafra et al. 2008)
According to several research findings, the number of antecedent dry days is one of
the most influential factors for build-up. A study carried out by Egodawatta et al.
(2006) in a residential landuse, Gold Coast, Queensland, Australia found that the rate
of build-up was initially in the range of 1 to 2 g/m2/day and decreased when the
31
antecedent dry days increased. Furthermore, they noted that particulate pollutant
composition varied dynamically when the antecedent dry days increased. They
suggested that this may be due to the re-distribution of fine particles by wind and
traffic. In the particle size distribution diagram (Figure 2.2) they developed, the
curve moves from left to right indicating the increase in the coarser fraction with
antecedent dry days. Furthermore, they noted that the average d50 values for 1 to 7
days were in the range of 75 to 100 µm, whereas it was around 200 µm for 14 days
and 250 µm for 21 days. From these observations, they suggested that though the
increase in build-up is limited, the solids composition changes continuously by
accumulating coarser particles and re-distributing finer particles when the antecedent
dry period increases. Furthermore, they noted that during this process, a higher
fraction of finer particles are more likely to deposit outside the road surfaces where
the turbulence is minimal.
Figure 2.2- Particle size distribution diagram (Adapted from Egodawatta et al. 2006)
Several other researchers (for example, Vaze et al.2000; Zafra et al. 2008) have also
noted that pollutant build-up increased with the antecedent dry period. According to
Duncan (1995) build-up is mainly a dry weather process. Bannerman et al. (1983)
found that large amount of solids are accumulated on urban catchments as a result of
atmospheric dry deposition and they estimated that it was nearly 50 mg/m2/day for
the catchment they studied.
32
Several studies have indicated that build-up is affected by natural and vehicle
induced winds. Studies by Deletic and Orr (2005) and Sartor et al. (1974) suggested
that pollutant concentration near the kerb of a road is significantly high in
comparison to the concentration of pollutants near the centre of the road. As they
noted, this is due to the movement of pollutants due to natural and vehicle induced
winds. Sartor et al. (1974) noted that only 5.9% of the particles near the kerb are less
than 43 µm. Anon (1981) noted that the mean particle size of the pollutants which
are deposited on road surfaces is around 15 µm. Novotny et al. (1985) found that at
least 20 m/hr wind speed is required for appreciable pollution re-distribution to
occur.
Pollutant re-distribution is significantly affected by parameters related to traffic.
According to the study by Kim et al. (2006), deposition from automobile exhaust is
composed of dust sized particles (<60 µm), but it is not the only source of traffic
related pollution. Tyre wear, solids carried on tyres and vehicle bodies, wearing of
parts such as brake pads and loss of lubrication fluids add to the pollution input
attributed to traffic. Vaze and Chiew (2002) found that the pollutant build-up may
vary along the longitudinal direction of the road depending on the slope of the road
and traffic signals present.
However, according to several researchers, the surface condition has a significant
impact on pollutant build-up. For example Sartor et al. (1974) noted significant
variation of pollutant load in US roads due to changes in road surface conditions. Pitt
(1979) found that the loading of pollutants on roads immediately after cleaning by
sweeping or rain was substantial and dependent on the road surface condition, with
rougher surfaces having higher loads. Furthermore, they found that the rate of
particulate re-suspension from road surfaces increase when the surfaces are dirty
(cleaned infrequently) and varied widely for different road surface conditions.
Egodawatta et al. (2006) attributed the significantly less pollutant loads on
Australian road surfaces to the variability of regional and management factors.
Researchers have noted that pollutant loading varies with the landuse (Herngren
2005; Goonetilleke et al. 2009). Sartor and Boyd (1972) found that pollution
accumulation rates on street surfaces vary with different landuses as shown in Table
33
2.3. From their study they concluded that industrial areas had the highest loads and
accumulation rates, due to less sweeping, more unpaved areas, spillages from trucks
and breakup of roads. Residential areas had intermediate loads, whilst commercial
areas had the lowest loads, due to better road surfaces and more frequent street
sweeping.
Table 2.3- Pollutant loading rates of street surfaces for different landuses
(Adapted from Sartor and Boyd 1972)
Landuse Loading rate (T/km)
Commercial 0.08
Residential 0.34
Industrial 0.80
The composition and particle size distribution of pollutants accumulated on road
surfaces are major concerns in water quality monitoring studies. Sartor and Boyd
(1972) found that 50% of metals and one third to one half of nutrients are adsorbed
to the fine fraction. Shaheen (1975) noted that the bulk of the accumulated particles
are in the range of 500-2000 µm. Herngren et al. (2005) found that around 85% of
the solids belong to the finer particle size groups which were smaller than 75 µm, in
the industrial and residential roads which they studied.
Egodawatta et al. (2006) found that a high fraction of pollutants is associated with
the fine particle size ranges. Furthermore, they noted that relatively high amount of
dissolved organic carbon was present in build-up samples. Dissolved organic carbon
enhances the solubility of other pollutants such as heavy metals and hydrocarbons
thus increasing their bio-availability (Herngren 2005; Warren et al. 2003).
Additionally, Gromaire-Mertz et al. (1999) suggested that a high fraction of organic
carbon in residential road surfaces can be attributed to the presence of trees and
adjacent grassed areas. As hypothesised by Sartor et al. (1974), the high
degradability of organic carbon would be the primary reason for the presence of high
concentration of DOC.
34
Several researchers (for example, Duncan 1995; Sartor et al. 1974) have noted that
pollutants accumulation on road surfaces can be formulated mathematically using a
decreasing rate of increasing function. For example, Shaheen (1975) developed an
equation for the pollutant build-up rate on an urban catchment as follows:
bb0b Mkk
db
dM −= Equation 2.1
Where,
Mb = amount of pollutant per unit area on the catchment surface (kg/m2)
k0 = constant rate of pollutant deposition (kg/m2 .h)
kb = constant pollutant removal rate (h-1)
b = time (h)
As discussed above, even though a number of studies have been already carried out
to investigate pollutant build-up on road surfaces, limited research literature is
available on build-up on roof surfaces (for example, Egodawatta 2007; Egodawatta
and Goonetilleke 2008; Yaziz et al. 1989). According to available literature,
pollutant build-up on roof surfaces varies with weathering of roofing material, dry
deposition, surface characteristics such as slope of the roof and roughness (Berdahl
et al. 2008; Yaziz et al. 1989). According to Yaziz et al. (1989), due to the steeper
slope and smooth surface of roofs, larger particles may not remain on roof surfaces
for long time compared to road surfaces.
According to Kennedy and Gadd (2001), dry deposition is the process where
pollutants from the atmosphere settle via gravity or deposited by impact of wind on
the roof surfaces. Dry deposition is affected by the surrounding natural environment
and climatic condition. For example, pollutant build-up on roof surfaces at coastal
areas could contain major sea salt irons such as sodium, chloride and magnesium.
Furthermore, Van Metre and Mahler (2003) noted that the amount of pollutants on
roof surfaces could vary with surrounding landuse, roof set up and antecedent dry
period. As noted by Van Metre and Mahler (2003), the build-up on roof surfaces
varies in the range 0.16-1.2 g/m2 depending on the magnitude of the antecedent dry
days. This is further supported by Egodawatta (2007) who noted that rate of build-up
35
is significantly high up to around seven days and then reduces after that as the
antecedent dry period progresses.
Egodawatta (2007) who investigated pollutant build-up on both road and roof
surfaces noted that build-up on roof surfaces are significantly finer compared to the
build-up on road surfaces. This could be attributed to the fineness of the atmospheric
depositions on roof surfaces. Furthermore, he noted that limited re-distribution of
pollutants occur with time on roof surfaces compared to road surfaces. This could be
primarily attributed to the reduced influence of vehicular induced wind turbulence on
pollutant build-up on roof surfaces.
2.4.2 Pollutant wash-off
Wash-off is the process by which accumulated dry deposition is removed from
surfaces and incorporated into stormwater runoff during rain events. Firstly, the
catchment surface gets wet with the rainfall and either dissolves the pollutants
accumulated on it or detaches particles from the surface. Then, due to the impact
energy of the raindrops, the pollutants are detached from the surface and are
incorporated into the runoff (Bujon et al. 1992; Hijioka et al. 2001). Wash-off of
pollutants is dependent on its availability on the surface, the energy of the rain drops
to loosen the material and the capacity of the runoff to transport the loosened
material (Pitt et al. 2004).
Wash-off behaviour is influenced by factors such as storm and catchment
characteristics (Duncan 1995; Goonetilleke et al. 2009). According to Cordery
(1977) most of the pollutants are washed from urban catchments by the high
intensities of rainfall which occur during first 10-20 minutes of rainfall and only
minor amounts are removed by subsequent rain. Chui (1997) showed that event
mean concentrations of COD and TSS increases with increasing rainfall intensity.
They concluded that this is due to the fact that higher rainfall intensities have a
greater capacity to scour materials deposited on a surface. They concluded that for
storms with higher rainfall depth, the total amount of pollutant load washed off will
be larger. Neary et al. (2002) found that wash-off was affected by both antecedent
36
dry conditions and rainfall intensity. Chen and Barry (2006) assumed that pollutant
wash-off load is proportional to or dependent on the accumulated pollutant mass on
the catchment surface before a runoff event. In addition, they found that the
pollutant wash-off load is a direct function of runoff volume which can be expressed
as follows:
( )rw vkb e1Ml −−= Equation 2.2
Where,
l = mass of pollutant washed off per unit area per rainfall event (kg/m2)
Mb = amount of pollutant per unit area on the catchment surface (kg/m2)
vr = runoff event volume (mm)
kw = pollutant wash-off coefficient (mm-1)
Researchers have noted that pollutant wash-off load is highly influenced by activities
associated with the urban landuse such as construction activities, vehicular traffic
and industrial activities (Bannerman et al. 1993; Pitt 2004). Construction activity and
other forms of soil disturbance can have a significant effect on wash-off. Duncan
(1995) found that wash-off loads can increase by a factor of 100 or more by
construction activity or soil disturbance in the catchment. Sartor and Boyd (1972)
noted that street cleaning and sweeping activities have a significant effect on
pollutant wash-off. They noted that street sweeping can remove a relatively large
fraction of the coarse particles accumulated on street surfaces. Sutherland et al.
(1998) found that pollutant abatement techniques such as street sweeping can only
efficiently remove relatively large particles of the order of 250 µm and larger.
However, in general, rain events can wash-off only a fraction of the build-up
pollutants from catchment surfaces. Vaze and Chiew (2002) found that after a
rainfall of 39.4 mm, only 35% of total pollutants were washed off. The following
rainfall event of 4 mm reduced total pollutant load by 45%. Furthermore, based on
field measurements they have proposed two possible wash-off concepts as shown in
Figure 2.3. The concepts are termed source limiting and transport limiting.
According to the source limiting process, the surface pollutant loads build-up from
37
zero over the antecedent dry days. Then the available pollutant load is washed off
during a storm event. On the other hand, in the transport limiting process, storm
events remove only a fraction of the pollutant load and build-up occurs relatively
quickly to return the surface pollutant load back to the level before the storm.
Figure 2.3- Hypothetical representations of surface pollutant load over time (Adapted from Vaze and Chiew 2002)
As reported by numerous researchers, the first flush has been noted as an important
and distinctive phenomenon within pollutant wash-off. The first flush relates to the
initial portion of the runoff being more polluted than the remainder due to the
washout of deposited pollutants by rainfall. It has often been noted that the
concentration peak precedes the runoff peak (Deletic 1998; Duncan 1995).
According to the study carried out by Weeks (1981) it was noted that overall,
approximately 40% of pollutants were exported by the initial 35% of runoff.
Furthermore, for high intensity storm events, he noted that approximately 48% of
pollutant loads are exported during the initial 37% of the runoff (Weeks 1981). Ellis
(1991) refers to research (Geiger 1987; Thornton and Saul 1986) showing that most
non-soluble pollutants up to 65% are washed off with the first 50% of the runoff
volume. Furthermore, he noted that soluble pollutants tend to have significant
removal during the initial runoff. According to Cordery (1977), pollution
concentration generally decreases with duration of rainfall. Furthermore, it was
noted that the rate of transmission of pollutants is much more dependent on the rate
of flow than on the concentration with most of the pollutant load occurring in less
than 1 hr.
38
Several researchers have investigated first flush effect of different pollutants in urban
stormwater runoff (for example, Cordery 1977; Deletic 1998; Furumai et al. 2001;
Howell 1978; Yufen et al. 2008). According to the study carried out by Hoffman et
al. (1984), concentration of pollutants such as heavy metals, hydrocarbons and solids
is higher during the first flush of a storm event. Van Metre and Mahler (2003)
investigating suspended solids wash-off behaviour on roof surfaces, noted that
during the first part of the wash-off of roof surfaces was laden with dark in colour
and after around 2.6 mm of rain it was clear. From these observations Van Metre and
Mahler (2003) suggested that majority of solids particles that are easily mobilized
and washed off during rainfall are removed by the first 2.6 mm of rain or less.
Horner et al. (1979) who investigated first flush effect of suspended solids and
chemical oxygen demand (COD) in highway runoff found that concentrations of
solids and COD to be higher in both magnitude and fluctuation during the first 30 to
60 minutes of a runoff event as shown in Figure 2.4. The fluctuations of pollutant
concentrations could be attributed to the variation of rainfall runoff characteristics.
Furthermore, Balades et al. (1984) found that 80% of COD, TSS and Pb were
eliminated by the first 52% of runoff. According to Berretta et al. (2007), the first
flush effect for rainfall intensities ranging from 1.8 mm/hr to 14.6 mm/hr is
significant for total hydrocarbons. In their study they noted that, for 80% of the
monitored rain events, the first 30% of runoff volume consisted of 40% to 60% of
total hydrocarbons. Forster (1996) has noted a significant first flush effect for heavy
metals in roof runoff after investigating runoff generated from a set of rainfall
intensities ranging from 1.2 mm/hr to 20 mm/hr.
Figure 2.4- First flush for solids and COD (Adapted from Horner et al. 1979)
39
The first flush pattern is also related to the rainfall intensity, the hydrological
characteristics of the catchment and the temporal pattern of the storm (Ellis 1991;
Han et al. 2006). Higher intensity rains have a greater ability to detach pollutants
from ground surfaces and to move particles that often have pollutants adsorbed to
them, giving higher pollution concentrations in the runoff (Duncan 1995; Shigaki et
al. 2007). According to Tiefenthaler and Kenneth (2001), the magnitude of the first
flush effect varies between intensities. He noted that suspended solids
concentrations in the first flush ranging from 112 mg/L for a rainfall intensity of 25
mm/hr to 140 mg/L for a rainfall intensity of 6 mm/hr.
Pollutant concentration in first flush is affected by the length of the antecedent dry
period and the surface condition. Forster (1996) suggested that pollutant
concentration in the first flush of roof runoff is affected by the surface condition. He
noted that runoff from rough roof surfaces have low concentrations of pollutants.
This is due to the ability of retaining more pollutants on the rough roof surfaces
specially during low intensity rain events. Furthermore, the length of the antecedent
dry period has a significant effect on pollutant concentration in the first flush. The
longer the dry period, the higher the concentration of pollutants in the first flush.
Thomas and Greene (1993) noted increasing trend in concentration of suspended
solids, turbidity and conductivity with the increase of antecedent dry days while zinc
and nitrate did not show such a trend.
2.5 Summary
The above discussion summarises the important conclusions drawn from the review
of literature. The research findings are based on knowledge of hydrologic and water
quality changes due to urbanisation, key stormwater pollutant indicators and
pollutant processes.
Urbanisation and the consequent increase in impervious surfaces and changes in land
use have a significant impact on the urban water environment. Increase in volume
and rate of runoff and peak discharge and reduction in base flow are the main
hydrologic impacts of urbanisation. On the other hand, deterioration in quality of
40
receiving water bodies due to urban stormwater runoff is significant. Anthropogenic
activities due to urbanisation produce a wide range of pollutants to urban stormwater
runoff. They degrade the physical, chemical and microbiological quality of
stormwater runoff.
The main types of pollutants in urban stormwater runoff are suspended solids,
organic carbon, nutrients, heavy metals and hydrocarbons. Behaviour of these
pollutants in urban stormwater runoff is complex due to their chemical properties.
The pollutants in urban areas originate from different sources. Activities associated
with urbanisation create a variety of sources of pollutants. To assess the impact of
urban stormwater pollution it is essential to understand the main pollutant processes.
In this context, pollutant build-up and wash-off are important.
41
Chapter 3 - Mitigation Actions and Stormwater Quality Monitoring
3.1 Background
As described in Chapter 2, polluted stormwater runoff is a leading cause of the
degradation of receiving water quality. Consequently, significant emphasis has been
placed on management strategies to mitigate the urban stormwater pollution.
Research into impacts of urban stormwater pollutants and development of effective
stormwater pollution mitigation actions are important elements underpinning
effective stormwater management strategies (Wong 2001). However, this
management needs in-depth knowledge of runoff quality. This knowledge is required
to understand the effects of runoff on receiving water quality and to develop
appropriate mitigation actions (Barrett et al. 1998; Han et al. 2006). However,
effectiveness of the mitigation actions can be limited due to the lack of knowledge
on water quality parameters and pollutant processes. In this context, stormwater
quality monitoring programs play an important role (Martinez 2005).
The primary objective of a monitoring program is to obtain information necessary to
make sound management decisions. For example, a typical stormwater monitoring
program may identify problems in specific areas and determine which problems are
the most significant. Lee and Stenstorm (2005) questioned whether the resulting
database of stormwater quality monitoring programs are adequate for planners and
regulators to identify acute problems and improve long term water quality
management plans. Outcomes from these stormwater quality monitoring
programmes have had limited success in creating significant new knowledge (Lee
and Stenstorm 2005; US FHWA 2001). This is primarily due to the difficulties
which arise in planning and conducting stormwater quality monitoring programmes
(US FHWA 2001). For example, dealing with a large number of parameters to be
tested increases the cost and time associated with monitoring.
42
In this context, a convenient approach to evaluating the quality of water directly
from field-based measurements without having to carry out resource intensive
laboratory experiments is of crucial importance. Therefore, it is feasible to identify a
set of easy to measure surrogate parameters and their relationships with the other
water quality parameters (Kayhanian et al. 2007; Settle et al. 2007). This chapter
presents an outline of current stormwater quality mitigation actions and then
examines current issues related to stormwater quality monitoring programs. Finally,
the chapter presents the outcomes of past research studies which have focussed on
identifying surrogate water quality parameters.
3.2 Current stormwater quality mitigation actions
As discussed in Chapter 2, stormwater runoff pollution is one of the most significant
environmental issues in urban areas. Pollutant loads originating from urban
catchments is significantly higher when compared to rural catchments, leading to
adverse impacts on receiving water quality (House et al. 1993; Lee et al. 2007;
Novotny et al. 1985; Sartor et al. 1974). Therefore, regulatory authorities are
challenged to implement appropriate stormwater management strategies.
Primarily, the purpose of stormwater management is to prevent and mitigate the
impacts of stormwater runoff through appropriate stormwater treatment measures
(Eric and Strecker 2000). The use of stormwater control practices to manage the
quality and quantity of urban runoff has become wide spread in many countries (Sara
et al. 2002; Urbonas 2000; Wong 2001). According to Sara et al. (2002), the
practices that promote long-term success of a stormwater management scheme is
referred to either as Best Planning Practices (BPPs) or Best Management Practices
(BMPs). There are two types of BMPs; Non- structural BMPs and Structural BMPs.
Non-structural BMPs can be described as a group and is a set of practices and
institutional arrangements. These aim to institute good housekeeping measures that
reduce or prevent pollutant deposition in urban areas (Urbonas 2000). Some common
non-structural BMPs are:
43
• Environmental and urban development policy- Environmental and urban
development policy is required to encourage widespread adoption of
ecologically sustainable development practices. This includes the incorporation
of Water Sensitive Urban Design (WSUD) into the urban planning process.
Water Sensitive Urban Design (WSUD) is a philosophical approach to urban
planning and design that aims to minimise the hydrological impacts of urban
development on the surrounding environment.
• Environmental considerations on construction sites- Poor planning and
management of construction sites can severely deteriorate the quality of
stormwater runoff. Therefore, proper site management is a useful strategy to
minimise the generation of pollutants from construction activities.
• Education and staff training- Education programs including staff training
should be directed at all staff levels. Training should provide the necessary
tools/techniques to enable staff to plan for future activities such as approval,
construction, operation or maintenance activities.
• Community education programs- Community education programs addressing
stormwater management issues encourage change in social ‘norms’ and
behaviours. Individual changes in behaviour may collectively contribute to
reduce the impacts of urban development on stormwater. More importantly, an
informed community can place pressure on government and industry to be
responsible for impacts on stormwater.
• Enforcement programs- Financial penalties are potentially an effective
proscription to activities that result in the pollution of stormwater. Enforcement
programs are largely the responsibility of the environmental protection
authority and local government.
(Adapted from Sara et al. 2002)
Structural BMPs can be described as stormwater treatment measures that collect,
convey or detain stormwater to improve water quality and provide a reuse function.
44
They are expected to function unattended during a storm event and to provide the
necessary treatment (Urbonas 2000). Structural BMPs are designed to function
without human intervention at the time stormwater flow occurs. Common examples
of structural BMPs include:
• Diversion of runoff to garden beds;
• Constructed wetlands- Vegetated system with extended retention time;
• Rainwater tank/reuse schemes (ie. For garden watering, toilet flushing);
• Sedimentation tanks- Concrete structures containing appropriate depth of water
to facilitate the settling of suspended solids under quiescent conditions;
• Filter drains- Gravelled trench systems where stormwater can drain through the
gravel to be collected in a pipe; unplanted but host to algal growth;
• Porous pavements- Continuous surface with high voids content, porous blocks
or solid blocks with adjoining infiltration spaces; an associated reservoir
structure which provides storage. No geotextile liner is present and host to algal
growth;
• Percolation trench- A percolation trench is a rock filled trench that temporarily
stores stormwater and percolates it into the ground. A percolation trench
typically serves small impervious tributary areas of two hectares or less;
• Swale- Vegetated broad shallow channels for transporting stormwater;
• Retention ponds-Contain some water at all times and retains incoming
stormwater; frequently incorporates vegetated margins; and
• Detention basins- Dry most of the time and able to store rainwater during wet
conditions and often possess a grassed surface.
(Adapted from Sara et al. 2002, Scholes et al. 2007)
According to Maestri and Lord (1987), four structural BMPs are identified as cost-
effective best management practises (BMPs) for stormwater runoff treatment. They
are vegetative controls, wet detention basins, infiltration basins and wetlands.
Furthermore, in order to provide successful mitigation actions Urbonas (2000)
introduced four tasks to be used as a tool to develop appropriate mitigation actions.
They are:
45
1. Prevention- Practices that prevent the deposition of pollutants on urban
surfaces.
2. Source control- Preventing pollutants from coming into contact with
stormwater runoff.
3. Source disposal and treatment- Reduction in the volume and/or rate of surface
runoff and the associated pollutant loads or concentrations at, or near their
source.
4. Follow-up treatment- Interception of runoff downstream of all source and on-
site controls using structural BMPs to provide follow-up flow management
and/or water quality treatment.
However, effectiveness of these tasks is still yet to be investigated. Several
researchers have questioned the significance of the implementation of a number of
these tasks in a particular mitigation action or use of a set of BMPs in various
combinations (for example, Schueler et al.1991; Urbonas and Stahre 1993).
Furthermore, several researchers (Arcy and Frost 2001; Sara et al. 2002), have
questioned whether the use of structural and non- structural BMPs together with
mitigation actions have the ability to achieve successful application of which they
are intended. Cave and Roesner (1994) estimated that typical non-structural BMPs
are likely to result in stormwater pollutant reductions of approximately 5%-10%,
while structural measures may reduce some stormwater pollutants by 50%-90%.
Therefore, it is often necessary to use a combination of both structural and non-
structural BMPs to achieve the desired water quality outcomes. This is known as the
treatment train approach (Schueler et al.1991; Urbonas and Stahre 1993; Urbonas
2000).
Pollutant removal mechanisms associated with BMPs involve physical, biological
and chemical processes. Physical processes primarily involve trapping gross
pollutants and coarse solids and sedimentation of finer silts and clay sized particles.
Once gross pollutants and coarse solids are removed, other pollutant removal
mechanisms namely, biological and chemical processes can be effectively applied
(Sara et al. 2002).
46
However, according to several research findings, effectiveness of the mitigation
actions are still limited (Fletcher et al. 2004; Scholes et al. 2007; Urbonas 2000).
Urbonas (2000) noted that many factors influence the effective performance of
structural BMPs. For example, the technology underpinning these practices is
currently under developed. Therefore, many controls are used without understanding
their limitations and effectiveness under field conditions. Sometimes, the operation
of these controls opposes regulatory expectations or academic predictions or beliefs.
Importantly, selection, design and construction and use are further complicated by
the stochastic nature of stormwater runoff and its variability with location and
climate. Consequently, the limited knowledge relating to pollutant build-up and
wash-off processes severely impact on the effectiveness of the mitigation actions.
With the absence of this knowledge, the effectiveness of stormwater mitigation
actions which focus on managing stormwater quality impacts has not yet been
wholly successful (Fletcher et al. 2004; Goonetilleke et al. 2005; Urbonas 2000). In
this regard, stormwater quality monitoring plays an important role.
3.3 Stormwater quality monitoring and issues
According to several research findings, successful stormwater quality monitoring
programs require well designed and controlled field studies over a number of years
(Schueler et al.1991; Urbonas and Stahre 1993; Urbonas 2000). The water quality
data generated from these monitoring programs will help to improve the
understanding of specific physico-chemical processes and interactions that govern
the transformation of pollutants in stormwater. In turn, this knowledge is essential
for the implementation of effective BMPs (Urbonas 2000). Scholes et al. (2007) have
produced a ranked list of BMP pollutant removal efficiencies based on different
pollutant removal capabilities of BMPs using limited monitoring data available for
the catchment they studied.
According to Grayson et al. (1997) and Eyre and Pepperell (1999), catchment scale
water quality monitoring programs can be classified into three categories as routine,
event sampling and spatially intensive. Routine monitoring involves the periodic
collection of samples. This includes collection of samples on a fortnightly, monthly
47
or yearly basis within the catchment (Macdonald et al. 1995). This approach is costly
and not much successful in identifying the exact causes of poor water quality. This is
due to the limited number of sample locations which produce a low number of
samples. Event sampling is the flow weighted collection of samples at a limited
number of sample sites which are typically located at catchment outlet (Kronvang
1992). The spatially intensive approach involves the collection of samples from a
large number of sites over a short period of time. This provides detailed information
from and across the catchment that can be used to assess the quality of water.
Though, this method is applied widely in monitoring projects in Australia, the cost
associated is much higher than other methods (Grayson et al. 1997).
Currently there is an increasing emphasis on implementing successful stormwater
quality monitoring programs (Fletcher et al. 2004). In this context, data collection
during a comprehensive stormwater quality monitoring program should comply with
comprehensive quality assurance and quality review of factors such as sampling
methods and review of analytical methods. Monitoring results could then be used to
develop control strategies, prepare plans and budget estimates for addressing the
problems related to polluted stormwater runoff (US FHWA 2001). According to Lee
et al. (2007), the overall goals of stormwater monitoring programs include the
identification of high-risk pollutants and its sources, the identification of total
maximum daily loads of pollutants and ultimately the reduction of stormwater
pollution. Consequently, most monitoring programs are designed with the following
objectives:
• Review water assessments to understand local problems;
• Develop source area monitoring to identify critical sources;
• Conduct treatability tests to verify performance of stormwater controls for local
conditions; and
• Assessment monitoring to verify the success of local stormwater management
approaches.
(Martinez 2005).
However, many constraints arise regarding the quality of data obtained from
monitoring programs, thus impeding the effectiveness of stormwater quality
48
monitoring programs. This is mainly due to the unpredictable behaviour of
stormwater quality due to the highly variable nature of stormwater runoff between
storms and during a single storm event. Consequently, a small number of samples is
not likely to provide a reliable indication of stormwater quality at a given site or the
effectiveness of a given BMP. Furthermore, it is difficult to determine BMP
efficiency with high levels of statistical confidence. Therefore, it is essential to
monitor stormwater runoff at a number of strategically located monitoring stations to
characterize stormwater quality over a large area. In this context, the collection of
data which accurately represents the spatial and temporal behaviour of pollutant
indicators is a significant challenge for any monitoring program (Grayson et al.
1996).
The concentration of pollutants in stormwater runoff is likely to vary significantly
over the period of a given storm event. Some of this variability can be captured
through the collection of multiple samples. Continuous records of pollutant
concentrations are required in order to have an in-depth knowledge of runoff quality
such as timing and magnitude of the variations in pollutant concentrations.
Unfortunately, acquiring such data sets need continuous measurements and are
resource intensive (US FHWA 2001).
A lack of information on monitoring areas is another concern in monitoring
programs (Martinez 2005). A wide range of parameters can potentially affect the
quality of stormwater discharges including geographic location, climatic conditions,
ecologic conditions, hydrologic conditions and landuse (US FHWA 2001). In
general, the availability of data records relating to these parameters is insufficient to
achieve targeted outcomes of monitoring programs. Furthermore, Cordery (1976)
noted that it is very rarely that comparisons are made regarding the quality of water
prior to and after urbanisation. They suggested that this may be due to the lack of
related data which affect stormwater discharges such as percentage of
imperviousness prior and after urbanisation. Martinez (2005) noted that limited
information is available regarding how the impervious areas are connected to the
drainage system which is one of the most important factors affecting urban
hydrologic analyses.
49
Method of sampling is another important issue which leads to deficiencies in
conducting stormwater quality monitoring programs (Grayson et al. 1996; Martinez
2005). Sample collection methods that fail to produce representative samples are
often significant sources of error in water quality data (Martin et al. 1992). Mainly
there are two types of sampling methods. These are, grab sampling and automatic
sampling.
Grab sampling is a manual procedure which involves the collection of discrete
samples from a waterway (see Figure 3.1). This sampling method is subject to the
influence of human error. Furthermore, according to Settle and Goonetilleke (2001),
strict collection protocols are needed in grab sampling to avoid variations in samples
caused by differences in the water column.
Figure 3.1- Grab (manual) sampling Adapted from Chrystal (2006)
Therefore, automatic sampling of stormwater has become a more viable method in
current stormwater quality monitoring programmes (Chrystal 2006; US EPA 2002).
Automatic samplers (see Figure 3.2) are recommended for large sampling programs
when better representations of flows are needed. Most importantly, automatic
samplers are more reliable than grab sampling (Martinez 2005). Automatic samplers
are installed to collect runoff at defined points within the flow and under all stream
event conditions. Therefore, information is provided from only one location within
the flow profile. While a discrete quantity of samples is used to define the overall
properties of the flow at a particular time or runoff condition, this creates some risk
50
of sample bias due to variations in concentration effected by spatial and temporal
variability of stormwater quality (Harmel et al. 2006; US EPA 2002).
Frequency of sampling is another important issue in stormwater quality monitoring
(Grayson et al. 1996; Harmel et al. 2006). Grayson et al. (1996) noted that sampling
frequency is usually very low in most monitoring studies. They noted that the
frequency of sampling is important essentially for discharge measurements from
structures such as weirs and flumes. Continuous measurements of discharge from
such structures are difficult specially from high flow events. Consequently, they
argued that discharge is not a good predictor of pollutants concentration as sampling
strategies for measuring discharges from high flow events are very poor. They
further suggested that this can be overcome by measuring the concentration of the
parameters of interest continuously and combining this with discharge measurements
to compute continuous load. However, cost effective methods for the continuous
measurement of pollutant concentrations are not yet available.
51
Figure 3.2- Automatic sampler
Personnel requirements for monitoring programs are also a major concern.
Committed, on-call field staff is essential for successful water quality sampling
projects. Field personnel should be well-trained on QA/QC (Quality Assurance and
Quality Control) methodology, equipment operation, basic hydrology and safety
considerations (US EPA 1997). Whether samples are collected manually or
automatically, personnel must make frequent trips to sampling sites to collect data
and retrieve water samples. In either case, field staff must also commit adequate time
to conduct necessary equipment inspection, maintenance and repair. Excessive delay
52
in these activities can result in changes in the chemical composition of water samples
and thus inaccurate representation of actual water quality (Harmel et al. 2006).
Testing of a large number of samples in laboratories is another issue. For example, a
typical automatic sampler installation will have 24 containers capable of collecting a
number of discrete samples and which can be replaced throughout extended rainfall
event conditions. This potentially will entail a large amount of laboratory testing. On
the other hand, during a comprehensive stormwater quality monitoring program, the
sites are monitored for a range of physico-chemical water quality parameters
including on-site measurements as well as laboratory tests for various parameters.
The numbers of samples that can be collected and analysed by a laboratory in a
reasonable time frame is determined by QA/QC guidelines and should remain within
the laboratory budget (Chrystal 2006; Harmel et al. 2006; US EPA 2002; US FHWA
2001; US EPA 2005). For example, Lee and Stenstorm (2005) found that testing of
heavy metals increases the laboratory costs double or triple in comparison to other
parameters. Consequently, conducting monitoring programs can be time consuming
and resource intensive based on the number of monitoring stations and range of
parameters evaluated. Furthermore, expert knowledge is required in selecting
appropriate test methods (Martinez 2005; Milne 2002; US FHWA 2001).
The efficiency and effectiveness of stormwater quality monitoring programs are
highly dependent on the above issues (US FHWA 2001). These issues can impede
sound management decisions relating to urban stormwater quality. Consequently,
there is an ever growing demand to increase the efficiency and effectiveness of
stormwater quality monitoring programs. In this context, identification of a set of
easy to measure parameters which can act as surrogates for other water quality
parameters is of crucial importance.
3.4 Surrogate water quality parameters
According to several research findings, the identification of a set of surrogate
parameters which are simpler or less expensive to measure provide a technique to
estimate certain water quality parameters (Gippel 1995; Grayson et al. 1996; Settle et
53
al. 2007). This is achieved by using a secondary parameter which has a close
relationship to the parameter of interest. These related or surrogate parameters are
statistically correlated to the more complicated or expensive parameters. According
to Gippel (1995) and Grayson et al. (1996) some of these relationships may exist
over only part of the measurement range and may occur only under specific physical
and environmental conditions. Therefore, they suggested that these relationships
have the potential for being unique to specific catchments, regions or event
conditions. However, regardless of these limitations, researchers have noted that
surrogate parameter relationships can be commonly used to understand urban
stormwater quality (Kayhanian et al. 2007; Settle et al. 2007; Thomson et al. 1997).
Researchers have noted that, total suspended solids (TSS), total dissolved solids
(TDS), total organic carbon (TOC), dissolved organic carbon (DOC), electrical
conductivity (EC) and turbidity (TTU) can be used as surrogate indicators for other
water quality parameters (Han et al. 2006; Kayhanian et al. 2007; Settle et al. 2007;
Settle and Goonetilleke 2001). According to several researchers, TSS can be
considered as an appropriate surrogate indicator for urban stormwater quality
(Gippel 2005; Sartor and Boyd 1972; Vaze and Chiew 2002; Zeng and Rasmussen
2005). This approach is all the more relevant as pollutants such as hydrocarbons,
heavy metals and nutrients are heavily bound to suspended solids (Atasoy et al.
2006; Ongley et al. 1981; Sartor et al. 1974; Urbonas 1994). Due to this reason,
suspended solids have been used as an indicator to measure other pollutants (Atasoy
et al. 2006; Urbonas 1994). The capacity for suspended solids to adsorb other
pollutants is influenced by the particle size, particle structure and physico-chemical
properties such as pH, electrical conductivity and organic carbon concentration
(Pechacek 1994; Warren et al. 2003).
Particle size distribution of suspended solids is a particularly important parameter as
it determines mobility of the particles and their associated pollutant concentrations
(Sansalone et al. 1998; Sartor and Boyd 1972; Vaze and Chiew 2004). According to
several researchers relatively higher amounts of pollutants are associated with finer
particle sizes (Goonetilleke et al. 2009; Herngren et al. 2005; Vaze and Chiew 2004).
Fine particles are able to adsorb higher concentrations of pollutants because they
have a relatively larger surface area per unit mass and therefore a higher adsorption
54
capacity than larger particles. Additionally, Beckwith et al. (1984) and Andral et al.
(1999) noted that a greater proportion of organic and clay materials contained in fine
particles of solids leads to adsorption of other pollutants. Sartor and Boyd (1972)
reported that over 50% of metals were found sorbed to solids smaller than 43 µm for
the road surfaces they investigated in residential, commercial and industrial landuses
in USA.
According to Sartor and Boyd (1972), even though solids smaller than 43 µm
fraction contain only 5.9% of the total solids by mass, they account for a high
percentage of nutrients, heavy metals and pesticides as illustrated in Table 3.1. A
similar result was obtained by Bradford (1977) who found that the fine fraction of
street dust accounted for approximately 6% of the total mass of solids and greater
than 60% of the trace metals. According to Arntson et al. (1985), suspended solids
are associated with potentially toxic metals including As, Cu and Pb. Latimer (1984)
found that, better correlations between suspended solids and Fe, Cu and Pb, because,
these metals are mostly associated with particulate matter in urban runoff.
Furthermore, they suggested that both heavy metals and hydrocarbons are more
heavily adsorbed by the fine particles because of the high electrostatic charge on the
particle surface. According to the study carried out by Herngren et al. (2005), TSS
provides a good indication of metals such as Fe, Al and Pb in different particle size
fractions. As Herngren et al. (2005) noted, Pb, Fe, Al were correlated with TSS and
the majority of these metals were in the size range 0.45 µm -75 µm.
55
Table 3.1- Fraction of pollutants associated with different particle size ranges- percentage by weight (Adapted from Sartor and Boyd 1972)
Parameter >2000 µm
840- 2000 µm
246-
840 µm
104-
246 µm
43-
104 µm <43 µm
Total Solids
24.4
7.6 24.6 27.8 9.7 5.9
Volatile Solids
11.0
17.4 12.0 16.1 17.9 25.6
BOD
7.4
20.1 15.7 15.2 17.3 24.3
COD
2.4
4.5 13.0 12.4 45.0 22.7
Kjeldahl Nitrogen
9.9
11.6 20.0 20.2 19.6 18.7
Nitrates
8.6
6.5 7.9 16.7 28.4 31.9
Phosphates
0
0.9 6.9 6.4 29.6 56.2
Total
Heavy metals 16.3 17.5 14.9 23.5 27.8
Total pesticides
0 16.0 26.5 25.8 31.7
According to several research findings (for example, Deletic et al. 1998; Grayson et
al. 1996; Lewis 1996; Packman et al. 1999; Settle et al. 2007) turbidity is a potential
surrogate measurement for TSS due to the strong correlation between turbidity and
TSS. However, some researchers have noted that the relationship between turbidity
and suspended solids concentration is largely confounded by variation in particle
size, particle composition and water colour (Gipple 1996; Packman et al. 1999).
According to Gipple (1996), variations in particle size can cause the turbidity to vary
by a factor of four for the same concentration of suspended solids. He noted that this
is due to turbidity instruments that are most sensitive to dispersions with particle
sizes of median diameter of 1.2-1.4 µm. Furthermore, Packman et al. (1999) noted
that these variations significantly affect the turbidity measurements depending on the
form of the relationship. They developed a relationship between turbidity and TSS
which was in logarithmic form and noted that even slight changes in TSS
concentrations have large effects on turbidity readings. However, the use of turbidity
as a surrogate parameter for TSS should be considered as this relationship can be
56
used for continuous measurement, which in turn can overcome the problem of
infrequent sampling of solids loads (Gipple 1996).
Several researchers have investigated the correlation of organic matter to other
pollutants (for example, Kronvang 1992; Ujevic et al. 2000; Wang et al. 2001).
Kronvang (1992) found a correlation between particulate organic matter and
particulate organic phosphorus. According to Warren et al. (2003), dissolved organic
carbon (DOC) increases the solubility of polycyclic aromatic hydrocarbons and
heavy metals that are attached to the suspended solids present in stormwater runoff.
The increase in solubility increases the soluble fraction of pollutants. Therefore, the
amount of bioavailable pollutants also increases with the increase in DOC.
Furthermore, organic carbon can influence the concentration of pollutants attached to
suspended solids. According to Wang et al. (2001) organic matter plays an important
role in the adsorption of PAHs in suspended solids. Ujevic et al. (2000) noted that,
the finer fraction of solids contains the highest concentration of organic matter and
heavy metals suggesting a correlation between organic matter and heavy metals.
Furthermore, a study carried out by Herngren et al. (2005), noted the influence of
DOC in the distribution of heavy metals in different particle size ranges of
suspended solids in urban stormwater runoff. According to them, the measurement
of DOC provides important information on the solubility of Zn and Cu available in
the stormwater.
According to several researchers (for example, Kayhanian et al. 2007; Settle et al.
2007) in order to substitute one surrogate water quality parameter for another, it is
essential to develop appropriate mathematical relationships between them. Settle et
al. (2007) developed a set of linear regression relationships between solids and
phosphorus parameters and their surrogate parameters. Several other researchers also
noted that linear regression relationships provide easy measurement of parameters of
interest in comparison to the power and logarithmic form of equations (for example,
Robien et al. 1997; Thomson et al. 1997). According to Kayhanian et al. (2007),
these relationships may be developed under site-specific or regional basis conditions.
It is possible that a shorter list of water quality parameters may serve as potential
surrogates for a larger list.
57
According to Thomson et al. (1997) the degree to which the developed surrogate
parameter relationships are applicable or its portability is also important. One
possible situation is that pollutant concentrations may not be dependent on the
physical characteristics of the selected area which would then allow for the
portability of the developed relationships. The developed relationships can then be
applied to any site and can be considered portable.
In a event, where the coefficients for the developed models in each site are identical,
the model would be perfectly portable between the sites investigated. On the other
hand, numerical changes in model coefficients are applicable for areas with different
physical and/or natural characteristics. In this case the model coefficients may be
related to physical characteristics of respective areas. However, it is important to
determine the nature of the portability of the developed surrogate parameter
relationships (Thomson et al. 1997). Furthermore, Settle et al. (2007) noted that the
relationships which they developed for phosphorus with its surrogate parameters are
considered to be unique to individual waterways and in particular catchment and
stream conditions. They suggested that the relationships which they developed may
be affected by variations within catchment characteristics such as landuse and soil
type. Therefore, the evaluation of the robustness of the relationships will therefore
require ongoing review.
3.5 Summary
Regulatory authorities commonly implement mitigation actions to safeguard urban
stormwater quality. These are mainly in the form of Best Management Practices
(BMPs), namely and non-structural BMPs and structural BMPs. Non-structural
BMPs can be described as a set of practices and institutional arrangements which
aims to institute good housekeeping measures. Wetlands, swales, retention basins are
commonly used structural BMPs. The effectiveness of these mitigations actions are
still limited due to the scarcity of knowledge on pollutant build-up and wash-off
processes. In this context, stormwater quality monitoring plays an important role.
58
Difficulties associated with stormwater quality monitoring such as costly laboratory
experiments, the highly variable nature of stormwater runoff and deficiency of
sampling methods impede the efficiency of stormwater quality monitoring. This in
turn limits the reliability of monitoring data which is essential for implementing
successful mitigation strategies. Consequently, there is an ever growing demand to
identify easy to measure surrogate parameters for other water quality parameters.
It is possible that Total suspended solids (TSS), total dissolved solids (TDS), total
organic carbon (TOC), dissolved organic carbon (DOC) and electrical conductivity
(EC) and turbidity (TTU) can be used as surrogate indicators for other water quality
parameters such as phosphorus and heavy metals. Identification of a set of surrogate
parameters for other water quality parameters will reduce the cost and time
associated with lengthy laboratory experiments.
59
Chapter 4 - Research Tools
4.1 Background
According to the research literature, an adequate knowledge of pollutant build-up
and wash-off processes is required to assess the impact of stormwater pollution on
receiving water bodies and to design methods for minimising these impacts.
Numerous research studies have been carried out to investigate the pollutant build-up
and wash-off processes on urban catchment surfaces (Sartor and Boyd 1972; Vaze
and Chiew 2002). In this regard, techniques such as vacuuming, sweeping and
brushing for pollutant build-up sampling and the use of natural rainfall and simulated
rainfall for pollutant wash-off sampling have been commonly adopted in stormwater
quality research studies (Egodawatta et al. 2006; Herngren et al. 2005; Vaze and
Chiew 2002).
As noted by researchers, urban impervious surfaces have a significant impact on
urban stormwater quality (Ball et al. 1998; Chang et al. 2004; Forster 1996; Forster
1999; Hoffman et al. 1985; Quek and Forster 1993). Consequently, two types of
impervious surfaces have been investigated in this research study. They are road
surfaces and roof surfaces. In regard to roof surfaces, two model roof surfaces were
used for the investigations. This was in order to eliminate safety issues associated
with the investigations on actual roof surfaces.
This chapter presents the tools and techniques which were used in the research study.
The selection of the research tools for field investigations was done after careful
consideration of factors such as ease of operation and maintenance and portability of
the apparatus in the field. Furthermore, the successful application of the selected
research tools in previous research studies carried out by Egodawatta et al. (2006)
and Herngren et .al (2005) were also taken into consideration. The tools which were
selected for the field investigations include:
60
• Vacuum collection system for collection of pollutant build-up and wash-off
sampling;
• Rainfall simulator for creating pollutant wash-off; and
• Model roofs for roof surface investigations.
Furthermore, this chapter describes the various multivariate analytical techniques
which were used for the data analysis. These analytical techniques were selected
after careful consideration of the data analysis requirements. The analytical
techniques which were used for this research include:
• Principal Component Analysis (PCA);
• Partial least Square (PLS); and
• Multi criteria decision making methods (PROMETHEE and GAIA).
4.2 Vacuum collection system
Several researchers have found that collection of pollutant build-up samples from
urban impervious surfaces can be carried out by vacuuming, sweeping or brushing
the surfaces (Chang et al. 2005; Gulson et al. 1995; Vaze and Chiew 2002). Brushing
and sweeping of surfaces are generally efficient in collecting coarse particles,
whereas vacuuming is more efficient in collecting fine particles (Bris et al. 1999;
Vaze and Chiew 2002). However, the combination of these techniques could further
enhance the collection efficiency of build-up from the surfaces. Furthermore, several
researchers have used vacuuming as a preferable technique for the collection of
wash-off samples (Egodawatta 2007; Herngren et al. 2005). Therefore, a specially
designed vacuum system which is a combination of brushing and vacuuming was
used for the build-up and wash-off sample collection in the research undertaken.
4.2.1 Selection of vacuum system
In past research studies, both industrial and domestic type vacuum systems have
been used for the collection of pollutants from urban impervious surfaces (Shaheen
61
1975; Tai 1991; Vaze and Chiew 2002). Vaze and Chiew (2002) used an industrial
vacuuming system to collect pollutants on urban street surfaces. According to their
study, the sampling efficiency of the system was due to the high power generated by
the system. According to Tai (1991), the retention efficiency of a conventional
domestic vacuum system which was used for his study was 96.4% for particles <75
µm. He suggested that this was due to the effective filtration system used in the
domestic vacuum systems.
Therefore, when selecting a vacuum system, it was obvious that power generation
and an efficient filtration system are two important factors which have to be
considered. These factors vary with the design of the vacuum system including
differences in parameters such as weight, ease of operation and number of filtration
elements (Saulius et al. 2001). The selection of the vacuum system which was used
for the research was based on the above factors.
The vacuum system selected was a Delonghi Aqualand model which consists of a
highly compact 1500 W motor and an efficient filtration system. The same vacuum
system was successfully used by Herngren (2005) and Egodawatta (2007) during
their research studies. The selected vacuum system is a simple portable system and it
can be used for large numbers of sample collections during field investigations
(Egodawatta 2007; Herngren 2005).
According to the research literature, a high percentage of primary stormwater
pollutants are attached to the finer fraction of solids (particles smaller than 150 µm)
(for example, Goonetilleke et al. 2009; Herngren et al. 2006; Vaze and Chiew 2004).
Therefore, the ability of the vacuum system to collect the finer fraction of solids is
important. The collection of finer particles was enhanced by an attachment which
contained a vacuum foot with a brush. The brush attached to the vacuum foot
dislodged the finer particles from the surface. Furthermore, the use of a vacuum foot
in the system concentrated the air flow into a smaller area so that the power of the
system is more effectively used to collect both fine and coarse particles that are
attached to the surface.
62
The vacuum system incorporated a water filtration technique as well as a High
Efficiency Particulate Air (HEPA) filter to ensure minimal escape of finer particles
through the exhaust system. The water filter technique used by the vacuum system is
illustrated in Figure 4.1.
Figure 4.1- The water filter system of Delonghi Aqualand model
According to the manufacturer’s specifications, the Delonghi Aqualand model
HEPA filter has a 99.97% efficient filtration level. The mechanism of the filter is to
direct the air intake through a column of water so that particulate pollutants are
retained in water. The pollutant sample retained in water can be easily extracted to
sample bottles for further analysis.
4.2.2 Sampling efficiency
The sampling efficiency of the vacuum system was tested under simulated field
conditions before actual field investigations. For this purpose, an area of 400 mm x
400 mm was selected from a road surface. The selected surface was coarse textured
with asphalt paving similar to the conditions of the actual surfaces in the field as
described in the Section 5.3. Figure 4.2a, 4.2b shows the selected sample of the road
63
surface and a section of the actual road surface that was used for the field
investigations.
Figure 4.2a- Section of sample Figure 4.2b- Section of surface
road surface investigated at Ceil Circuit
A known weight of 100 g of soil sample which was selected to represent a pollutant
sample and analysed for particle size distribution. The pollutant sample represented
the particle size range of 1 to 1000 µm and this is in the generally expected range of
particle size distribution of pollutants on road surfaces. Prior to testing, the surface
was cleaned by repeated vacuuming and flushing with water and it was allowed to
dry by applying a stream of air. Then the graded pollutant sample was distributed
over the surface using a straight edged and a fine brush. During this work special
attention was taken to prevent any spilling of solids over the test plot.
After cleaning the vacuum cleaner compartment, hoses and foot thoroughly, the
vacuum system was filled with 3 L of deionised water. The soil sample, which was
spread on the sample surface, was collected by the vacuum system by vacuuming
three times in a perpendicular direction under simulated field test conditions. Then
the vacuum cleaner compartment was emptied to a clean container and washed
thoroughly to ensure all the pollutants collected were transferred into the container.
Further, all the hoses and the brush were washed thoroughly and the water was
poured into the same container in order to ensure that all the vacuumed particulate
pollutants were collected.
The collected sample was oven dried and the recovered solids were weighed. The
total sample recovery efficiency was found to be 95%, which could be considered
adequate for field investigations. Additionally, the recovered sample was analysed
64
for particle size distribution for comparison with the original sample and the results
obtained are shown in the Figure 4.3.
0
5
10
15
20
25
30
35
40
45
0 200 400 600 800 1000 1200 1400Particle size (µm)
Wei
ght r
etai
ned
(g)
Original sample
Vaccumed sample
Figure 4.3- Comparison of particle size distribution of original sample and recovered samples
The particle recovery efficiency for each particle size class was greater than 88%.
The loss of particles in each particle size class of recovered sample could be
attributed to systematic errors in sample filtration and the entrapment of particles in
the vacuum cleaner compartment and hoses. Therefore, it was important to wash the
vacuum cleaner compartment and hoses thoroughly to ensure that no particles
remained in order to minimise the losses in the field investigations. However, the
percentage loss is expected to be minimal in the field as the investigations were
subjected to larger area.
4.3 Rainfall simulator
Wash-off was created using simulated rain events. According to research literature,
the use of natural rainfall and simulated rainfall are two common approaches adopted
for pollutant wash-off studies (Goonetilleke et al. 2005; Herngren et al. 2005;
65
Shigaki et al. 2007; Vaze and Chiew 2002). However, the variability associated with
rainfall characteristics such as intensity, duration and kinetic energy in natural
rainfall has constrained the transferability of research outcomes which need to
generate the fundamental knowledge on the pollutant wash-off process (Herngren et
al. 2005). Furthermore, the random nature of occurrence of natural rainfall events
makes the investigations more difficult. In this context, the use of simulated rainfall
has become a common practice in water quality research studies due to the reduction
in the number of variables which constrain the transferability of research outcomes
(Egodawatta et al. 2006; Herngren et. al 2005). Consequently, the wash-off data
collection in this research study was based on a range of simulated rainfall events for
which a rainfall simulator was used.
The rainfall simulator (Figure 4.4), which was used for this research, was designed
and fabricated by Herngren (2005). According to Herngren et al. (2005), it was
designed to comply with the following requirements:
• Complete portability, ease of assembly and operation during use;
• Produce drop size distribution, terminal velocity and kinetic energy similar to
natural rainfall;
• Ability to create proposed rainfall intensities which are suitable for the research;
and
• Provide a satisfactory system for collecting runoff from impervious surfaces.
The simulator consisted of an A-frame structure made of Aluminium tubing of 40
mm diameter with three Veejet 80100 nozzles. The nozzles were mounted equally
spaced on a swinging nozzle boom. The nozzle boom was connected to a small
motor in order to swing in either direction. This arrangement allowed water to be
spread uniformly in either direction. As shown in Figure 4.4, two catch trays were
connected to aluminium tubes that were located under the nozzle boom. The water
return system of the simulator is controlled by these catch trays.
66
Figure 4.4- Schematic diagram of the rainfall simulator used for the study
(Adapted from Herngren et al. 2005)
The speed of the swing and delay time is controlled using an electronic control box.
Prior to use, the control box settings were calibrated for different rainfall intensities
which were selected for the study. The water was pumped to the simulator using an
externally located tank, such that the water pressure of the nozzle boom could be
adjusted to achieve the required drop size distribution and velocity. This was to
ensure satisfactory replication of natural rainfall characteristics. The runoff
collection system was designed for a 2 m x 1.5 m plot area which was connected to a
collection trough made from sheet metal and ensured no leakage of runoff from the
plot. More details of the arrangement of the rainfall simulator can be found in
Herngren (2005).
4.3.1 Calibration of the rainfall simulator
The main purpose of the rainfall simulator is to replicate natural rainfall events. In
this context, the most important rainfall characteristics to consider are drop size
67
distribution, rainfall intensity and kinetic energy of the rainfall. In order to provide a
reliable data base, an accurate replication of these characteristics is essential during
the simulation. Therefore, the calibration of characteristics of simulated rain events
is important prior to the investigations.
The rainfall simulator which was used for this research study was first calibrated for
its intensities and verified for kinetic energy and drop size distribution by Herngren
(2005). The experimental procedure which he had used is well explained in the thesis
of Herngren (2005). However, after the repeated use of the simulator over several
research studies, it was noted that the simulator did not operate properly due to the
wear and tear of the mechanical components in the simulator. Therefore, prior to
using the simulator for this research study, it was re-calibrated for the six different
selected rainfall intensities and verified for drop size and kinetic energy. The
selected rainfall intensities were 20 mm/hr, 40 mm/hr, 65 mm/hr, 86 mm/hr, 115
mm/hr and 135 mm/hr. More details on the selection of rainfall intensities can be
found in Chapter 6. The following sections describe the calibration procedure for the
rainfall simulator for the selected rainfall intensities and verification for drop size
and kinetic energy.
4.3.2 Calibration for rainfall intensity and uniformity of rainfall
The calibration of the rainfall simulator for rainfall intensity and uniformity of
rainfall was carried out similar to the procedure used by Herngren (2005) and
Egodawatta (2007). Firstly, twenty containers were placed under a plot area of 2 m x
1.5 m in a grid pattern as shown in Figure 4.5. This was to measure an average depth
of water collected for a known simulated rain duration. Secondly, the control box of
the simulator was set to a known setting. The control box consists of two types of
controls. One is to control the speed of oscillation and is demarcated as 1 to 5 on the
control box. The second is to control the delay time and is demarcated from A to M.
At the specified settings of the control box, simulated rain was generated for a 5
minute duration. The amount of water collected in the containers was then measured
and converted to depth of water per unit of time (mm/hr). The same procedure was
68
repeated for different settings of the control box and the complete set of data
generated is tabulated in Appendix A, Table 1. It was noted that the simulator was
capable of simulating rainfall intensities ranging from 20 mm/hr to 160 mm/hr.
Therefore, the use of the rainfall simulator to replicate the selected rainfall intensities
given in Section 4.3.1 was considered satisfactory. The control box settings that were
relevant to simulating the selected rainfall intensities in this research are given in
Table 4.1.
Figure 4.5- Arrangement of rainfall simulator for the intensity calibration and uniformity testing of rainfall simulator
Table 4.1- Selected control box setting for different rainfall intensities
Rainfall intensity
(mm/hr)
Speed setting Delay setting
20 1 A 40 1 H 65 2 J 86 3 K 115 6 L 135 4 M
In order to ensure a uniform distribution of rainfall over the area, the spatial variation
of the rainfall intensity was considered. Spatial variation of rainfall intensity
69
corresponds to the mean rainfall intensity at steady state for a particular rainfall
event (de Lima et al. 2003). This can be evaluated by the use of a uniformity
coefficient (Rickson 2001). The uniformity coefficient was calculated using the
collected data during the rainfall intensity calibration according to Equation 4.1 as
given below (Christiansen 1942):
( )nm
X1100Cu
×−
= ∑ Equation 4.1
Where,
Cu = Coefficient of uniformity
X = Absolute deviation of individual observation from mean value
m = Mean value
n = Number of observations
The uniformity coefficient is expressed as a percentage and more uniform the rainfall
intensity throughout the plot, the more the uniformity coefficient approaches 100%.
The uniformity coefficient obtained for the different rainfall intensities tested in this
study was around 70%. This was considered sufficient for a successful rainfall
simulation (Herngren 2005; Loch et al. 2001).
4.3.3 Drop size distribution and kinetic energy of rainfall
According to researchers, the drop size distribution and kinetic energy of rainfall are
two important parameters which should be considered in order to ensure the
capability of the rainfall simulator to achieve a better replication of rainfall events
(Herngren et al. 2005; Loch 1982). Kinetic energy in an individual rain drop is
greatly influenced by the drop size. This is due to the variation of both mass and
terminal velocity of the drop. As the drop size varies with the rainfall intensity,
kinetic energy is in turn, affected by the rainfall intensity. Several researchers (for
example; Herngren 2005; Loch et al. 2001; Roswell 1986) have noted that, for
rainfall intensities of 0 to 40 mm/hr, the kinetic energy varies between 0 to around 25
J/m2/mm.
70
Therefore, the rainfall simulator was originally designed to simulate kinetic energy
in this constant region where rainfall intensities are greater than 40 mm/hr.
According to Herngren (2005), this was done by adjusting the pressure at the nozzle
boom to 41kPa. During rainfall simulation, different rainfall intensities were
simulated by varying the nozzle boom’s movement. Since there is no change in
simulator hydraulics, kinetic energy was constant for all of the rainfall intensities.
Therefore, verification testing was carried out only to check the potential of the
simulator for producing the initially calibrated drop size and kinetic energy under a
pressure of 41 kPa.
According to research literature, a number of methods have been used for the direct
measurement of rain drop size distribution (Assouline et al. 1997; Hudson 1963).
The two most widely used methods are the stain method and the flour pellet method.
In the former method, drops are allowed to fall on a uniformly absorbent surface
such as blotting paper. The latter method uses pellet making media, such as flour or
cement. In both methods, the drop size is obtained by comparing the size of stains or
pellets with those produced by the drops of a known diameter (Assouline et al. 1997;
Bubenzer et al. 1984; Hall et al. 1970; Herngren 2005; Hudson 1963). Compared to
the stain method, the preparation of the experimental setup for the flour pellet
method was easier as it had fewer technical difficulties. Therefore, the flour pellet
method was used for this research study.
The flour pellet method was initially developed by Hudson (1963). This method was
further recommended through, its use in the research of Herngren (2005) and
Egodawatta (2007). In this method, an uncompacted layer of flour is exposed to
simulated rain for a few seconds and a number of rain drops are allowed to fall on
the flour. Then the flour is oven dried for 12 hours at 1050C and the resultant pellets
are passed through a set of sieves and separated into the following different particle
size ranges (Figure 4.6).
• >4.75 mm;
• 4.75 mm - 3.35 mm;
71
• 3.35 mm - 2.36 mm;
• 2.36 mm - 1.68 mm;
• 1.68 mm - 1.18 mm;
• 1.18 mm - 0.85 mm; and
• <0.85 mm.
Figure 4.6-Pellets separated into each size ranges The average weight of a pellet was calculated by dividing the weight of the total
amount of pellets by the number of pellets available in each size class. The next step
was to calculate the drop mass which formed the pellet. For this, a calibration curve
developed by Hudson (1963) was used. The calibration curve developed by Hudson
(1963) represents the relationship between the pellet’s mass vs. the ratio of drop
mass with the pellet’s mass However, the direct use of this calibration curve could
not be recommended for this research because of influential factors such as the type
of flour used and the degree of compaction of the flour. The variations in these
factors can cause errors in the results. Therefore, in order to eliminate the effect of
these factors, the calibration curve of Hudson (1963) was validated. A pilot
experiment was conducted to determine the applicability of using the Hudson (1963)
curve in calculating the drop mass of the simulated rain events.
In the set up used for the pilot experiment, a medical needle with a known diameter
was connected to a large reservoir which had been placed at a height of 3 m, so that
72
water droplets released from the needles reached a velocity close to their terminal
velocity during the fall (Figure 4.7). A known number of drops was collected to a pre
weighed beaker in order to calculate the average drop weight. The beaker was lined
with cotton wool to avoid splashing and evaporation. Flour pellets were made
simultaneously by replacing the beaker with a tray containing a thick layer of
uncompacted flour and a known number of pellets was made and oven dried.
The separated and cleaned flour pellets were weighed to determine the average
weight. The experiment was repeated for ten needles with different diameters and the
data points obtained are plotted in Figure 4.8. From this experiment, it was not
possible to obtain smaller sizes of pellets. Therefore, the pilot experiment was able to
verify only a range of the calibration curve of Hudson (1963). However, the results
obtained were in close agreement with the calibration curve. Therefore, it suggested
that Hudson’s (1963) calibration curve and procedure could be used to calculate the
raindrop sizes.
3m
Figure 4.7- Experimental setup for drop size calibration
Reservoir
Needle
Cotton wool
Collection beaker
73
0.5
0.7
0.9
1.1
1.3
1.5
0.1 1 10 100Pellet Mass (mg)
Mas
s R
atio
(Dro
p M
ass/
Pel
let M
ass)
Hudson, (1963)
Experiment
Figure 4.8- Calibration curve for flour pellets As the average weight of a pellet was already calculated, it was converted to drop
mass by using this calibration curve. The drop mass was then converted to the drop
diameter and the median drop size (D50) was calculated. The median drop size (D50)
represents the complete drop size distribution of a particular storm event. According
to Rickson (2001), the median drop size (D50) is defined as the drop size where 50%
of drops generated in the storm are larger and 50% are smaller. The calculated D50
for this study was 2.45 mm. According to Hudson (1963), the median drop size for a
natural rainfall event is in between 2 mm to 2.5 mm. Therefore, the calculated drop
size was accepted for this research. More details on the calculation of drop size is
given in Appendix A, Table 2.
The terminal velocity for each drop size class was estimated based on Laws (1941)
data. According to Herngren et al. (2005), the simulator height of 2.4 m was
sufficient for creating terminal velocities similar to natural rainfall for all drop sizes.
Therefore, it was assumed that all the drops reached the terminal velocity. Using the
average drop size diameter and the corresponding terminal velocity, the kinetic
energy of rainfall was then calculated. This was done by taking the sum of kinetic
energy of individual drops. The calculated value of the kinetic energy was 25.63
J/m2/mm. This value was in close agreement to the kinetic energy obtained by
Herngren et al. (2005) for the same simulator. Therefore, the values obtained for
74
drop size and kinetic energy ensured a satisfactory replication of natural rainfall
kinetic energy.
20 mm/hr was the lowest intensity rainfall event selected for simulation in the
research study undertaken. According to Rosewell (1986), the typical kinetic energy
for 20 mm/hr rainfall intensity is in the range of 16 to 18 J/m2/mm. However, it was
noted that the simulator is only capable of simulating rainfall with a constant kinetic
energy of 25.63 J/m2/mm. Therefore, in order to simulate 20 mm/hr intensity, the
kinetic energy of the raindrops needed to be reduced accordingly. This was done by
using a kinetic energy dissipater introduced by Egodawatta (2007). In accordance
with this method, a meshed fly screen frame was placed just below the nozzles in
order reduce the drop size by breaking the large droplets. In turn, the reduction in
drop size reduces the mass and terminal velocity of raindrops and leads to a
reduction in kinetic energy. Egodawatta (2007) confirmed the suitability of this
method in reducing the kinetic energy to simulate 20 mm/hr rainfall intensity by re-
calibrating the simulator used in this study.
4.4 Model roofs
As discussed in Chapter 2, roof surfaces are identified as significant contributors of
pollutants to urban stormwater runoff. Since roof surfaces represent a relatively
higher fraction of impervious surfaces in urban catchments, they generate higher
volumes of stormwater runoff with a significant amount of pollutants (Bannerman et
al. 1993; Chang and Crowley 1993; Egodawatta 2007; Van and Mahler 2003).
According to Chang and Crowley (1993), more than 50% of impervious surfaces in
residential catchments are represented by roof surfaces. Pollutants generated from
sources such as dry atmospheric deposition, wet deposition and weathering of
roofing materials are accumulated on roof surfaces and wash-off with the roof runoff
(Kennedy and Gadd 2001). Consequently, an investigation of pollutant build-up and
wash-off characteristics on roof surfaces is similarly important to that of road
surfaces. This will lead to effective stormwater quality mitigation strategies.
75
However, an investigation of pollutant build-up and wash-off processes on actual
roof surfaces is not an easy task. This is mainly due to safety issues and difficulties
in obtaining permission from residents and authorised organisations, such as the city
council. Furthermore, technical difficulties can arise when using the rainfall
simulator for investigations on actual roof surfaces. Therefore, two model roof
surfaces were used for roof surface investigations. These model surfaces were
initially designed and fabricated by Egodawatta (2007).
As the test area for the rainfall simulator was 1.5 m × 2.0 m, the roof surfaces had
also been designed for the same dimensions. The model roofs were made from two
different types of roofing materials, corrugated steel and concrete tiles. These
roofing materials are the most widely used roofing materials in South East
Queensland, where the study sites were located. Figure 4.9 shows the model roof
surfaces used for this study.
The roofs had been designed at an angle of 200 based on the guidelines provide by
the roof material manufacturers. Furthermore, roof surfaces consist of scissor lifting
arrangements. Using this arrangement, roof surfaces can be lifted to a typical single
storey roofing height for pollutant accumulation and then lowered to ground level
sample collection. A hydraulic jack powered by a 2.4 kW hydraulic pump was used
to lift the roof surface. The experimental set up and procedure adopted is well
documented in the thesis of Egodawatta (2007).
76
Figure 4.9- Model roof surfaces used in the study
4.5 Data analytical tools
In general, water quality research studies require large data bases with a range of
parameters. The resulting large data matrix needs to be carefully analysed in order to
produce substantive outcomes. These outcomes are essential for extending the
knowledge base on pollutant processes and key water quality parameters. This
knowledge will eventually help in the development of appropriate mitigation actions
to safeguard the urban stormwater quality. In this context, the application of both
univariate and multivariate (chemometrics) statistical data analysis techniques has
become a valuable tool in water quality research studies (Adams 1995; Goonetilleke
et al. 2005; Vega et. al 1998).
In this research, a number of pollutant build-up and wash-off samples from road and
roof surfaces were tested for a range of physico-chemical parameters which are
important in terms of urban stormwater quality. Consequently, this generated a wide
array of data. The data analysis techniques were selected carefully by considering the
type of data to be analysed, the capabilities of different data analysis techniques and
the type of analysis to be performed. These techniques were selected to understand
77
the pattern recognition of the data variability, to asses the relationships between
objects and variables and for predictive purposes.
4.5.1 Univariate data analysis techniques
Univariate analysis of a given data set explores each variable in the data set
separately. It looks at the range of values, as well as the central tendency of the
values. It describes the pattern of response to each variable individually. Hence, prior
to focusing on multivariate data analysis, univariate statistical analysis techniques
were used to understand the primary variability of the physico-chemical parameters
investigated in the research.
Mean and standard deviation (SD) are two univariate statistical measurements that
are widely used to describe the characteristics of a single variable data set (Adams
1995; Bahar et al. 2007; Goonetilleke et al. 2005). Mean is the ordinary arithmetic
average of the data set. However, extreme values or outliers in the data set affects the
mean. Therefore, for a non normal data distribution, the mean value is not a good
summary statistic.
Standard deviation (SD) measures the dispersion of data about the mean value and
measurement in the same units as the data. A large standard deviation indicates that
the data are scattered widely about the mean value and conversely, a small standard
deviation is characteristic of a more tightly grouped set of data (Adams 1995).
4.5.2 Multivariate data analysis techniques
The focus of this research was to identify a set of easy to measure surrogate
parameters. Therefore, it was needed to identity linkage between parameters. As the
univariate data analysis is based on comparing only two parameters, it was difficult
to use only the univariate data analysis techniques in order to identify linkage
between multiple number of parameters. This was overcome by the application of
multivariate data analysis techniques. These techniques are useful for understanding
78
the relationships between variables when a large number of data is available. For
multivariate data analysis, variables are not considered in isolation but are combined
to provide as a complete description of the total data set as possible (Adams 1995).
The multivariate data analysis techniques which were used in this research study are:
1. Principal Component Analysis (PCA);
2. Partial Least Square Regression (PLS); and
3. Multi Criteria Decision Making Methods - PROMETHEE and GAIA.
1. Principal Component Analysis PCA)
PCA has been widely used as a pattern recognition technique in numerous water
quality research studies to analyse multivariate statistical data (Alberto et al. 2001;
Goonetilleke et al. 2005; Kokot et al. 1998; Librando et al. 1995; Settle et al. 2007;
Vega et. al 1998). For example, Vega et. al (1998) used PCA to produce a
meaningful classification of river water samples affected by seasonal influences.
Settle et al. (2007) used PCA for the investigation of physical and chemical
behaviour of solids and phosphorus in urban stormwater runoff. Herngren et al.
(2005) used PCA to understand correlations between physico-chemical parameters
such as total organic carbon, total suspended solids and heavy metals in different
particle size fractions of solids wash-off in three different landuses.
PCA reduces the dimensionality of the data set by explaining the correlation among
the large set of variables in terms of a small number of underlying factors or
principal components. It facilitates the extraction of information about the
relationships among objects and variables in a data matrix (Settle et al. 2007).
Further, once this task has been achieved, data are presented diagrammatically.
Therefore, this is highly recommended in research studies as researchers are very
responsive to pictorial presentations. Additionally, PCA is useful to observe the
relationship between objects and the variables together on the same diagram (Kokot
et al. 1998).
PCA operates mathematically from the covariance matrix, which describes the
dispersion of the multiple measured parameters, to obtain eigenvalues and
79
eigenvectors. Linear combinations of the original variables and eigenvectors result in
new variables, which are known as principal components (PCs) (Alberto et al. 2001).
PCs are orthogonal components and they are uncorrelated to each other.
Furthermore, these PCs lie along the direction of maximum variance of the data.
Therefore, the transformation is achieved without loss of information of the data set.
The first PC contains most of the data variance and the second PC contains the
second largest variance and so on. The analysis can produce the same number of PCs
as the original data set, but the first few PCs contain most of the variance. Therefore,
the first few PCs are selected for interpretation (Adams 1995; Jackson 1991; Kokot
et al. 1998).
The number of PCs, which is selected for interpretation is typically determined by
using the scree plot method (Jackson 1991). Scree plot indicates the variation of the
eigen values in a descending order with corresponding principal components. The
number of principal components is determined according to the point where the
graph first shows a significant change in the slope (Adams 1995). To apply this
method, the data must be arranged into a data matrix with columns defining the
selected variables and the rows referring to the sample measurements. In order to
avoid any influence due to the magnitude of the variables, the data are subjected to
standard pre-treatment techniques. The most common pre-treatment techniques are
standardisation, mean centering and auto scaling (Kokot and Yang 1995; Libarando
et al. 1995; Nguyen et al. 1999; Tyler et al. 2007). In a given data matrix,
standardisation means that each cell in a given column is divided by the standard
deviation of that particular column. Thus, each variable will now be equal in
weighting with a standard deviation of 1 (Kokot et al. 1998). Mean centering
consists of subtracting the mean value of each variable from each element in their
respective column. In PCA, mean centered data tends to describe the first PC in the
direction of the largest variance in the data (Kokot and Yang 1995). Auto scaling is
the combination of both standardisation and mean centering.
Finally, the pre-treated data set is subjected to PCA. Each PC generated can be
characterised as loadings and scores. The scores describe differences or similarities
between samples or objects. The score value on each PC is the projection of the
object on to the given PC. The loadings explain the variation in the scores. Loading
80
refers to the weighting of variables on each PC. The contribution of variables to each
PC is displayed by positive and negative values of weights. High positive or negative
values indicate the important variables for the relevant PC and low loading values
reflect the unimportant variables. However, these low variables could be of
significance on another PC. The loading plot is complementary to the scores plot and
it is an essential tool in interpreting which variables are responsible for the patterns
observed on the PC scores plot. The relationships between the objects and the
variables are often best displayed on a biplot. This is a plot with axes so scaled as to
include the scores-scores coordinates as well as the loading values (Kokot et al.
1998; Pommer et al. 2004).
In the PCA biplot, vectors representing parameters which are close together were
taken as correlated parameters whereas an obtuse angle indicates a weak correlation.
Parameters those which are at an angle of 90º angle were taken as uncorrelated
parameters. However, the PCA biplot alone was not adequate to identify the linkage
between parameters specially when the biplots explain low total data variance.
Therefore, the correlation matrix which shows the degree of correlation among the
parameters was also used in the analysis in order to identify the best correlated
parameters. When the best correlated parameters are identified, it can be decided the
potential surrogate indicators for a parameter of interest.
2. Partial Least Square Regression (PLS)
PLS is a well known factor analysis method which is principally applied for
prediction. PLS provides the ability to construct predictive models when the factors
are many and highly collinear (correlated) (Randall 2003).Therefore, it has become a
standard tool for modelling relationships between multivariate data (Ayoko et al.
2007; Purcell et al. 2005). Herngren (2005) used the PLS approach to establish
predictive relationships for PAHs and heavy metals in particulate and dissolved
fractions of sediments in urban stormwater runoff. Ayoko et al. (2007) applied PLS
to model and predict water quality parameters using a set of data obtained from
investigating the surface water and ground water in selected developing countries.
From the results, they identified a set of water quality parameters which can be used
to predict the physico-chemical characteristics of surface and ground water. Einax et
81
al. (1998) used PLS for the evaluation and interpretation of river pollution data.
Aguilera et al. (2000) used PLS to assess the quality of the coastal water in tourist
areas in Spain.
PLS is a regression extension of PCA, combining the strengths of PCA and common
least square regression (Eriksson et al. 2001). PLS works with two matrices referred
to as X and Y. PLS regression is a recently developed generalisation of multiple
linear regressions (MLR). According to Wold et al. (2001) MLR works well as long
as the X-variables are fairly few and fairly uncorrelated. On the other hand, PLS
regression is of particular importance because unlike MLR, it can analyse data with
strongly correlated, noisy and numerous X-variables (predictors). Furthermore, it can
be also simultaneously model several Y-response variables.
In the application of PLS, the original variables are summarised by the calculation of
new variables, called latent variables. The latent variables are linear combinations of
the original variables. They are orthogonal to each other. Furthermore, these latent
variables are interpretable and rich in practical information (Einax et al. 1998; Sun
2004). In the PLS modelling, both matrices are decomposed into a matrix of latent
vectors and a loading matrix plus a residual matrix. This is undertaken under the
condition which maintains a maximum correlation between both matrices of the
latent vectors (Einax et al. 1998). Hence, in the modelling both X and Y data are
actively used in the data analysis (Zhang et al. 2006).
Generally, developing a PLS model is a two stage process, namely, model
calibration and model validation. Therefore, firstly, for the application of PLS, data
matrices should be divided into two sets with one for calibration and the other for
validation. This is done according to the split rule, where, one half of the data matrix
is used for calibration and the remainder is used for validation.
To develop the PLS model, it is needed to decide on the number of principal
components. Principal components (PCs) are extracted in such a way that the first
PC carries most of the information, followed by the second PC and so on. However,
at a certain point, the variation modeled by any new PC can be mostly attributed to
82
noise. Therefore, it is essential to determine the correct complexity of the PLS
model. In this regard, different validation methods exist to select the optimal number
of PCs. In this research, the PLS model was developed according to leave one-
sample-out at a time, cross validation method. This method has been successfully
applied in water quality research studies in the past (Aguilera et al. 2000; Ayoko et
al. 2007).
Cross validation is a method where each sample in the calibration is predicted to give
an estimate of the prediction accuracy of the calibration. This generally gives an
over-optimistic idea of the actual performance of the model. The numbers of
principal components which are needed to develop the model were identified by
examining the decrease in the pattern of the standard error of cross validation
(SECV) plot which is a function of the error versus the number of principal
components (Kim et al. 2007). The error of the model is indicated by SECV value.
Furthermore, Regression coefficient (R2) is used to indicate the precision achieved in
calibration.
Both SECV and R2 are commonly used statistical parameters in determining the
predictive ability of a calibration. Furthermore, these give relative judgment of
model performance (Dunn et al. 2002). Low SECV with high R2 indicates the
excellent validity of the calibration model (Dunn et al. 2002; Faber and Kowalski
1997; Khalil 2004). More details of these statistical parameters are available
elsewhere (Faber and Kowalski 1997; Khalil 2004). The cross validation procedure
provides a reasonable first approximation of the predictive power of a model.
However, it is known that cross-validation sometimes produces over-optimistic
results when working with strongly correlated data (Eriksson et al. 2001).
In this context, in order to estimate the predictive power, the validity of the model
was further explored by undertaking external validation. Although the calibration set
provides a good indication of the suitability of the model when used for prediction,
the validation is also useful as it will allow more beneficial testing of the overall
efficiency of the model (Dunn et al. 2002; Goonetilleke and Thomas 2004). The data
83
set which is kept for validation is used when testing the predictive ability of each
calibration. The best calibration is the one with the highest coefficient of
determination (r2) and the lowest Standard Error of Performance (SEP). SEP
indicates the standard deviation in the difference between actual and predicted values
for samples in a validation set. It is important to consider SEP values in relation to
the laboratory error and how the data will be used in practice (Batten 1998).
PLS regression has two approaches. They are PLS-1 (unicomponent models) and
PLS-2 (uni and multicomponent models). PLS-1 is used to perform the
decomposition and regression for only one component at a time. PLS-2 is used to
calculate the loadings on the basis of all concentrations simultaneously (Falco et al.
2002; Ferrer et al. 1998; Nevado et al. 1998). According to Martens (2001), PLS-2 is
more helpful when many Y variables are available. Furthermore, PLS- 2 is faster and
slightly simpler to use than PLS-1. Moreover, PLS-2 maximises the covariance
between linear combinations of X and linear combinations of Y.
In past research studies, both of these methods had been used successfully. Falco et
al. (2002) used both of these approaches to determine the presence of Chromium and
Cobolt in water samples. Ferrer et al. (1998) used these methods to predict PAH
concentrations in water samples. In their study, they used both approaches and
suggested that PLS1 is more effective than PLS2 in predicting PAHs due to its
capability to treat each component individually.
3. Multi Criteria Decision Making Methods- PROMETHEE and GAIA
Multi Criteria Decision Making Methods (MCDM) facilitates decision-making when
dealing with multivariate problems. The main objective of MCDM is to help
decision-makers to solve complex decision problems in a systematic, consistent and
more productive way. According to research literature, numerous multicriteria
decision-making methods have been used successfully as a tool for decision-making.
SMART, ELECTRE, SMAA, PROMETHEE and GAIA are some of the most
common methods which have been used in previous research studies (Khalil et al.
2004; Lahdelma et al. 2003; Lim et al. 2006; Martin et al. 2007). Martin et al. (2007)
84
used the ELECTRE III method to assess the performance of several BMPs and to
select possible stormwater source control strategies for an urbanised area. All these
methods are designed to provide a decision by comparing the performance of, or
preference for one object to another.
However, among the recent application of MCDM methods, PROMETHEE
(Preference Ranking Organization Method for Enrichment Evaluation) and GAIA
(Graphical Analysis for Interactive Assistance) are identified as the two most
common ranking methods over the other methods (Ayoko et al. 2007; Brans et al.
1986; Carmody et al. 2005; Herngren et al. 2006; Khalil et al. 2004). According to
Brans et al. (1986), the PROMETHEE method is more suitable than ELCTRE due to
its simplicity, clearness and stability in application. Also, unlike PCA,
PROMETHEE has the ability to provide ranking even for a few objects.
Though the application of PROMETHEE and GAIA in water quality research has
not been extensive, but the methods have been successfully applied in several
research studies (Herngren et. al 2006; Khalil et al. 2004; Settle et al. 2007). Settle et
al. (2007) used PROMETHEE to rank the water samples collected from two
catchments in Brisbane, Queensland, Australia based on the concentration of
measured sets of physico-chemical parameters. Additionally, they used GAIA
analysis to identify the linkages between physico-chemical parameters and hence, to
identify potential surrogate parameters for suspended solids and total dissolved
phosphorus.
Herngren et al. (2006) used these methods to analyse the distribution of heavy metals
in different particle size ranges of road deposited solids at three different landuses in
Queensland State, Australia. In their study, PROMETHEE was applied to identify
the most polluted particle size range of the solids and the landuse in terms of heavy
metals. GAIA was used to identify the linkage between heavy metals and the
different particle size ranges of sediments. Additionally, they used GAIA to identify
possible relationships between heavy metals and total organic carbon content.
Khalil et al. (2004) used PROMETHEE and GAIA for site selection for sustainable
on-site sewage effluent disposal based on the physico-chemical characteristics of the
85
different types of soil. These methods were applied to assist in understanding the
relationships between different physico-chemical parameters and important soil
properties in the context of effluent renovation capacity. Ayoko et al. (2007) used
MCDM to predict the physico-chemical properties of surface and ground water in
Papua New Guinea. They used PROMETHEE and GAIA to rank the water bodies,
which were selected in terms of water quality and to find patterns in the parameters
that influence water quality.
In this research, PROMETHEE and GAIA were used to analyse the pollutant build-
up data for two main reasons. Firstly, due to the small number of build-up samples
obtained from the field investigations, it seemed more reliable to use PROMETHEE
and GAIA analysis as it provides good interpretation even with lower numbers of
objects. Secondly, PROMETHEE ranked the objects based on the pollutant loads
and this was needed to understand the primary characteristics of the build-up solids.
Decision Lab software was used for the PROMETHEE and GAIA analysis (Visual
Decision Inc. 2000). The underlying theory in relation to these methods has been
discussed in detail elsewhere (Carmody et al. 2005; Keller et al. 1991; Khalil et al.
2004). However, a summary description of PROMETHEE and GAIA methods is
given below.
PROMETHEE
PROMETHEE is a nonparametric method, which ranks a number of objects or
actions based on the criteria or variables in the data matrix. In this study, objects
refer to the build-up samples which have been collected from road and roof surfaces.
The criteria refer to the range of physico-chemical parameters, such as TSS and TOC
measured during the laboratory analysis. Prior to ranking or ordering of a number of
actions (objects), pre-selected (by user) ranking order, weighting condition,
preference function and threshold value were applied to the variables (Carmody et al.
2005; Keller et al. 1991; Khalil et al. 2004). A brief description of these values is
given below.
• Ranking Order (preferred ranking sense)
Minimised and maximised conditions are allocated for each criterion to establish the
preferred ranking sense. Minimised values imply the lower value of variables and
86
maximised values imply the higher value of variables. The ranking order of actions
can be undertaken from the bottom-up (minimised) or top-down (maximised)
depending on the decision-maker’s preference (Carmody et al. 2005; Khalil et al.
2004; Settle et al. 2007).
• Weighting
Weights can be allocated to each criterion to reflect the importance of one criterion
over another. The criterion weight is a positive value. It is independent of the scale
of the criterion. The higher values of weights give more importance to the criterion.
However, weights can be altered by the decision-maker if alternative scenarios are
required in the analysis. Otherwise, by default, a weighting of 1 can be assigned for
all criteria (Carmody et al. 2005; Visual Decision Inc. 2000).
• Preference Function
The preference function, P (a, b) provides the mathematical basis for selecting one
object in preference to another. Furthermore, it translates the deviation between the
evaluations of two samples on a single variable into a preference degree. The
preference degree represents an increasing function of deviation. Therefore, smaller
deviations will contribute to weaker degrees of preference and larger ones contribute
to stronger degrees of preference. In Decision Lab, six shapes of preference
functions are available and they are described in Table 4.2 (Carmody et al. 2005;
Visual Decision Inc 2000).
• Threshold
The shape of the preference function is dependent on the threshold value. For most
functions, one or two classification thresholds must be provided by the user. Q,
which is known as the indifference threshold represents the largest deviation. When
comparing two actions on a single criterion, this is considered to be negligible by the
decision-maker. Preference threshold, P represents the smallest deviation that is
considered decisive when comparing the two actions. P is always greater than Q. The
Gaussian threshold S is a middle value which is only used with the Gaussian
preference function (Keller et al. 1991; Khalil et al. 2004).
87
Table 4.2- List of preference functions (Adapted from Khalil et al. 2004)
Preference function Threshold Shape
Usual No threshold
U-shape Q threshold
V-shape P threshold
Level Q and P thresholds
Linear Q and P thresholds
Gaussian S threshold
A brief description of PROMETHEE procedure is given below.
Step 1.
This is a transformation of the raw data matrix to a difference matrix, d. This
includes that, for each criterion, all of the column entries(y) in the raw data matrix,
compared pairwise by subtracting all possible combinations, which give a difference,
d for each comparison.
Step 2.
For each criterion, the selected preference function, P (a, b) (Table 4.2) is applied to
determine how much the outcome a is preferred to b. The sum of preference values
for all criterions for each object gives a value called ‘global preference index’ (π). Π
indicates the preference of one object over another.
Step 3.
In this step, the results of all the comparisons done using the preference functions are
summarised. For this, (Φ+) and negative (Φ−) outranking flows are calculated by
88
summing up all the global preference indices. The positive outranking flow (Φ+),
indicates how an object outranks all others. The negative outranking flow (Φ−),
shows how all others outrank each object. A higher value for (Φ+) and a lower value
for (Φ−) indicate a higher preference for an object. Moreover, a net outranking flow
(Φ) can be calculated by taking the difference between (Φ+) and (Φ−) (Equation
4.2).
Φ(a) = Φ+(a) - Φ−(a) Equation 4.2
The larger the net flow, the better the object is considered relative to the other
objects.
Step 4.
This involves a comparison of positive and negative outranking flows to produce
partial pre-ordering (ranking) of the objects (actions). This is known as
PROMETHEE 1 and is based on three possible outcomes:
a. One action is preferred to another;
b. There is no difference between the two actions;
c. The two actions cannot be compared.
PROMETHEE 1, in the partial outranking graph (Figure 4.10) is based on the
intersection of the ranks induced by Φ+ and Φ−. According to this, as a rule,
comparable objects are joined by one or more arrows, while incomparable objects
are unconnected by arrows.
Figure 4.10- PROMETHEE 1: partial outranking graph
a4
a5
a6
a2
a1
a3
89
PROMETHEE 1 is a very useful tool to expose objects which cannot be compared.
Therefore, it is more informative.
Step 5.
A Complete ranking method, PROMETHEE II (Figure 4.11) is produced from the
net outranking flow (Φ). This eliminates the incomparability option (c) noted above.
Although it may be more convenient to use PROMETHEE II complete ranking,
some information does get lost in the process, which is retained in PROMETHEE 1,
partial ranking (Brans 2002; Carmody et al. 2005; Kokot and Phuong 1999; Purcell et
al. 2005; Settle et al. 2007).
Figure 4.11- PROMETHEE II: ranking
GAIA GAIA is a procedure for the display and evaluation of PROMETHEE results. The
GAIA matrix is constructed from a decomposition of the Φ net outranking flows.
The data are then processed by a PCA algorithm and displayed on a GAIA biplot.
Similar to the PCA procedures, GAIA reduces a large number of variables to a
smaller number of principal components and visually shows how the variables relate
to each other and objects. However, unlike other PCA results, GAIA displays a
decision axis, π. The decision axis π displays the degree of decision power pointing
to the approximate location of the preferred action. Furthermore, GAIA provides
some guidance for criteria, which are important for net outranking and determining
which criteria influence the decision axis (Carmody et al. 2005; Keller et al. 1991;
Purcell et al. 2005). The interpretation of the GAIA plot requires little elaboration, as
it is identical to the PCA biplot.
Interpretation of GAIA biplots are summarised as follows.
a4
a54
a6
a2
a1
a3
90
• Criteria (variables) are represented by vectors. Both the orientation and the
length of the vectors are important.
• The longer a projected vector for a variable, the more variance it contains.
• If the variable vectors are oriented in the same direction, they are correlated. i.e
the preferences are similar.
• Independent variables have almost orthogonal vectors, while equivalent
variables have close vectors and conflicting variables have vectors in the
opposite direction.
• Objects projected in the direction of a particular variable are strongly related to
that variable, while the opposite objects are weakly related to that variable.
• Similar objects are visualized as clusters and dissimilar objects can be found in
different directions, i.e. on different PC coordinates.
• If the decision axis, π is long, then the decision power is strong. If the vectors
are not too conflicting, the best choices are those that are the closest possible to
the decision axis, π, and farthest possible from the origin.
• If the decision axis, π is short, then the decision power is weak, vectors are
conflicting, the best choices are the closest possible to the origin as they do not
correspond to any extreme.
(Espinasse et al. 1997; Keller et al. 1991; Visual Decision Inc. 2000)
4.6 Summary
The main research apparatus which were used for this research were:
1) Vacuum collection system;
2) Rainfall simulator; and
3) Model roof surfaces.
The vacuum collection system was used to collect pollutant build-up and wash-off
samples from the surfaces. Prior to the use of the vacuum system in the field, the
system was tested for the particle collection and retention efficiency. Using the
selected vacuum system for field investigations in this research project was found to
be satisfactory.
91
A rainfall simulator was used to create wash-off data from the study surfaces. The
rainfall simulator was calibrated for six different rainfall intensities and verified for
drop size and kinetic energy. For the simulation of smaller rainfall intensities, a
kinetic energy dissipater was introduced.
Two model roofs were used for the investigation of roof surface pollutant build-up
and wash-off. The use of model roofs overcomes the difficulties of collecting build-
up and wash-off samples from actual roof surfaces. These model roofs consisted of a
scissor arrangement. By using this arrangement, roof surfaces could be lifted to the
necessary roofing height or lowered to ground level for build-up or wash-off
sampling.
Both univariate and multivariate data analysis techniques were used for the data
analysis. Univariate data analysis techniques were applied to explore the data set.
However, as larger numbers of variables were involved in the data set, multivariate
data analysis techniques were preferred. Consequently, principle component analysis
for pattern recognition and partial least square regression to establish predictive
relationships between variables, were employed. In addition, multi criteria decision
making methods, PROMETHEE and GAIA were used for ranking purposes and to
visually display the relationship between variables and objects in the analysis of
build-up data.
92
93
Chapter 5 - Selection of Study Sites
5.1 Background
The field investigation methodology for the research undertaken was developed to
obtain pollutant build-up and wash-off data to specifically identify a set of easy to
measure surrogate water quality parameters. Selection of field study sites was one of
the important components of the research methodology. The selection of study sites
was done after careful consideration of the suitability of the sites to conduct field
investigations.
Landuse was not considered to be an important consideration as the surrogate
parameter relationships were considered to be independent of landuse. As discussed
in Chapter 2, as all urban impervious surfaces have a profound impact on urban
stormwater quality, two types of impervious surfaces were selected for the field
investigations. The selected study sites represented both road and roof surfaces. This
chapter presents a detail description of the selected study area, selection of the study
sites and key criteria for selecting the sites.
5.2 Study area
Gold Coast was selected as the study area for field investigations. Gold Coast is the
sixth largest city in Australia. It is located just south of Brisbane, which is the capital
city of Queensland, Australia. The study sites are situated within the residential
suburb of Coomera. Coomera is located about 40 km south of Brisbane, at the
northern end of the Gold Coast (Figure 5.1). This region is one of Australia’s fastest
growing urban areas and is expected to grow from approximately 10,000 people to
around 120,000 people by 2025. The area provides a comfortable living to residents
in terms of environmental, social and aesthetic aspects (Coomera Waters Ltd. 2009).
94
Figure 5.1- Location of Coomera (Adapted from Google- Map data - 2009)
Gold Coast is one of the few places in Australia where a comprehensive urban
catchment monitoring program is being undertaken. The study sites which were
investigated were in a catchment which is currently being monitored for a range of
water quality parameters to assess the treatment efficiency of a range of water
sensitive urban design (WSUD) infrastructure to reduce the adverse impacts of
stormwater runoff. This monitoring study is being undertaken in collaboration by the
Department of Environment and Natural Resources, Gold Coast City Council and
Queensland University of Technology. These treatment devices being monitored
include a grass swale, bio-retention basin and a wetland (Figure 5.2).
95
Figure 5.2- Monitoring sites of Coomera (Adapted from Parker et al. 2008)
5.3 Study site selection
A high fraction of pollutants originate from urban impervious surfaces. Among
them, road and roof surfaces are of crucial importance as they have been recognised
as major contributors of pollutants to urban stormwater runoff (Bannerman et al.,
1993; Egodawatta, 2007; Huang et al. 2007; Jian-Wei et al. 2007; Van Metre and
Mahler, 2003;). Consequently, two road surfaces and two roof surfaces were selected
within the study area for the field investigations. The selection of study sites was
based on the following criteria:
• Convenient accessibility to the sites;
• Minimum disturbance to residents and traffic;
• Convenience in setting up the rainfall simulator;
• Sufficient slope for the flow of runoff;
• traffic conditions in the area; and
96
• Serviceability of the catchment.
5.3.1 Investigation of road surfaces
The two selected road surfaces were Drumbeat Street and Ceil Circuit (Figure 5.3
and Figure 5.4a, 5.4b). Both Drumbeat Street and Ceil Circuit are access roads with
detached family houses with small gardens. This is typical to urban residential
developments in the region. Additionally, the small gardens and lawns beside the
each road seem to be well turfed and maintained.
Figure 5.3- Location of study sites (Adapted from Google- Map data 2009)
Location of study sites
97
Figure 5.4a- Study site 1- Drumbeat Street
Figure 5.4b- Study site 2- Ceil Circuit
98
5.3.2 Investigation of roof surfaces
The roof surface investigations were conducted using two model roof surfaces.
Details of model roofs are discussed in Section 4.4. Considering the relevant safety
issues, it was decided to install them in a suitable place with a lock-up space and
restricted public access. The roof surfaces were installed in the same residential area
where the road surface investigations were undertaken (Figure 5.5a, 5.5b).
Figure 5.5a- Deployment of tile roof surface
99
Figure 5.5b- Deployment of steel roof surface
5.4 Summary
Coomera which is a suburb in Gold Coast was selected for the field investigations.
Two road surfaces and two sites for locating the roof surfaces were selected for the
investigations. Road surfaces were selected after the careful consideration of factors
such as road surface condition, easy accessibility to site, convenience in setting up
rainfall simulator and minimum disturbance to residents. Two model roof surfaces
were selected for the roof surface investigations after considering the difficulties
which arises in undertaking investigations on actual roof surfaces. They were placed
in the same area where the road sites were selected.
100
101
Chapter 6 - Sample collection and laboratory testing
6.1 Background
As discussed in Chapter 5, two road surfaces and two roof surfaces were selected for
the field investigations. This chapter presents the sample collection techniques used
in the field investigations and the analytical procedures adapted for laboratory testing
of collected samples. Sample collection during the field investigations were in two
stages; collection of pollutant build-up samples and wash-off samples. As discussed
in Chapter 4, a specially designed vacuum system and a rainfall simulator were used
to collect pollutant build-up samples and wash-off samples.
Based on the knowledge gained from the literature review, build-up and wash-off
samples were tested for the following physico-chemical parameters:
• pH;
• Electrical conductivity (EC);
• Turbidity (TTU);
• Particle size distribution;
• Total suspended solids (TSS);
• Total dissolved solids (TDS);
• Total organic carbon (TOC);
• Dissolved organic carbon (DOC);
• Different nitrogen compounds-Nitrite-nitrogen (NO2-), Nitrate-nitrogen (NO3
-),
Total kjeldahl nitrogen (TKN) and Total nitrogen (TN) and
• Different phosphorus compounds- Ortho-phosphates (PO4-3), Total phosphorus
(TP).
With reference to numerous published literature, these parameters are the key
indicators of stormwater quality (Ball et al. 2000; Sartor and Boyd 1972; Settle et al.
2007; Vaze and Chiew 2002). In addition these parameters have been widely used to
102
develop computer models which are essential for implementing mitigation actions to
safeguard the urban stormwater quality.
6.2 Collection of samples
6.2.1 Collection of pollutant build-up samples from road surfaces
Pollutant build-up on road surfaces depends on a range of factors such as landuse,
traffic characteristics and antecedent dry days (Bian and Zhu 2008; Pitt 1979; Sartor
and Boyd 1972). The road surfaces which were selected as study sites were in a
residential area. Therefore, landuse was not considered as a variable. The selected
road sites did not appear to have a significant variation in traffic flow. A minimum
seven day antecedent dry days was allowed prior to collection of samples in order to
allow sufficient time for pollutant build-up. This time period was based on the
findings of Egodawatta et al. (2006) who noted that the build-up load increases with
the antecedent dry days and approaches a near constant value after about seven days.
A plot of size 2.0 m × 1.5 m was selected at each road surface for the investigations.
The plot was selected equidistant from the road median and the kerb in order to
maintain the consistency of the build-up sampling. The boundary of the plot was
demarcated by a wooden frame. As discussed in Chapter 4, a specially modified
vacuum cleaner was used to collect the build-up samples (Figure 6.1). Before use, all
components of the vacuum cleaner were cleaned with deionised water. Additionally,
3 L of deionised water was added to the vacuum cleaner compartment as the
filtration agent.
The demarcated plot was vacuumed three times each in perpendicular directions in
order to ensure that all the available pollutants were collected. At the end of
vacuuming, the sample retained in the filtration compartment was transferred to a
polyethylene container. This polyethylene container was pre-washed according to
standard methods (APHA 2005). Finally, the vacuum compartment and all the hoses
103
were thoroughly washed using deionised water and the liquid was added to the
container.
Figure 6.1- Collection of pollutant build-up samples from road surfaces
6.2.2 Collection of pollutant wash-off samples from road surfaces
As discussed in Chapter 2, pollutants wash-off on road surfaces is affected by a
range of parameters such as rainfall intensity, duration and the pollutants load
accumulated on the surface (Chui 1997; Egodawatta 2007; Neary et al. 2002). In this
research, primary variables considered for wash-off investigations were rainfall
intensity and duration (Egodawatta et al. 2006; Yaziz et al. 1989). Consequently,
sample collection was carried out for six simulated rain events. Each intensity was
simulated only once for each study site. The selected rainfall intensities and their
durations are shown in Table 6.1. The selection of these rainfall intensities and
durations were based on regional rainfall events in the Gold Coast area and
represents more than 90% of the regional rainfall events (Egodawatta 2007).
104
Table 6.1- Rainfall intensities and durations simulated during the study (Adapted from Egodawatta 2007)
Rainfall Duration (min) Rainfall Intensity
(mm/hr) Event 1 Event 2 Event 3 Event 4
20
40
65
86
115
135
10
10
10
10
5
5
20
15
15
15
10
10
30
25
20
20
15
15
40
35
30
25
20
20
Wash-off investigations were conducted on the same road surfaces where the build-
up investigations were conducted. A stretch of roadway with average surface
condition was selected at each study site for rainfall simulations. As the amount of
pollutants wash-off from a surface is dependent on the amount of pollutant build-up,
the same size of plot area which the build-up sample was collected was used for
wash-off investigations. A 1.5 m × 2.0 m frame with rubber flaps was used to
demarcate an individual plot area (Figure 6.2) for the sample collection.
Similar to the build-up sampling plot, the wash-off sampling plot areas were also
selected equidistant from the road median strip and the kerb. It was assumed that the
amount of pollutants on the road surface was the same at each individual test plot for
a specific road site. Each rainfall intensity was simulated starting from the
downstream end of the selected road stretch and moving upstream for the next
simulated rainfall event. The runoff water from the rainfall simulation was collected
using a catch tray and vacuum system as shown in Figure 6.2.
Samples were collected in 5min time intervals for each rainfall intensity. The time
interval was selected for easy handling of runoff samples collected. As discussed in
Section 4.2, a specially designed vacuum system was used to collect the wash-off
samples. The wash-off samples were directed into clean 25 L polyethylene
containers at the same time of vacuuming (Figure 6.3).
105
Figure 6.2- Set-up of the rainfall simulator in the study site
Figure 6.3- Collection of samples to polyethylene containers
Set up of the rainfall simulator
Vacuum cleaner for sample collection
3m2 plot area
Catch tray
106
6.2.3 Collection of pollutant build-up samples from roof surfaces
As discussed in Chapter 4, two model roofs were used for the roof surface
investigations. Both roof surfaces were kept at one location in the same study area
which was used for road surface investigations. Samples were collected in three
sampling episodes representing antecedent dry days of 8, 6 and 6. Unlike road
surfaces, it was not possible to ensure a minimum 7 day antecedent dry period for the
second and third sampling episodes due to practical difficulties associated with
leaving the roofs in the study area.
On each roof surface, half the area (1.5 m2) was used to collect pollutant build-up
while the other half was used for wash-off sampling (See Figure 6.4). Altogether 6
build-up samples were collected from both roof surfaces. Build-up samples were
collected by washing the roof surface four times with 7 L of deionised water.
Additionally, a soft brush was used for brushing the surface. A common roof gutter
was placed to collect the sample and to direct it to a polyethylene container kept
underneath the gutter opening (See Figure 6.4). The gutter was thoroughly washed
before and after each sample collection.
Figure 6.4- Collection of pollutant build-up samples
107
6.2.4 Collection of pollutant wash-off samples from roof surfaces
Similar to wash-off sample collection from road surfaces as discussed in Section
6.2.2, wash-off sample collection from roof surfaces was conducted using the rainfall
simulator. The same rain events (See Table 6.1) as for road surfaces were simulated
on roof surfaces. Wash-off sampling was carried out on the remaining half of the
roof surface which was not used for build-up investigations. This was done by fixing
the gutter at the other half of the roof surface (See Figure 6.5).
A total of six intensities were simulated during the three sampling episodes
consisting of 65 and 86 mm/hr intensities during the first sampling episode, 115 and
135 mm/hr intensities during the second sampling episode and 20 and 40 mm/hr
intensities during the third sampling episode. 20, 86 and 135 mm/hr intensities were
simulated on the steel roof surface and remaining intensities namely, 40, 65 and115
mm/hr were simulated on the tile roof surface. For the simulations, the rainfall
simulator was placed exactly above the lowered model roof. The simulator was
raised to maintain 2.5 m average height from roof to nozzle boom of the simulator.
Unlike the wash-off investigations on road surfaces, it was not possible to consider 5
min time slots in the collection of samples from roof surfaces as runoff is generated
much faster on roof surfaces. Therefore, frequent sample collection was necessary
during the initial part of each event and simulations were conducted until relatively
clean runoff was observed. Samples were directed to the containers which were kept
underneath the gutter as shown in Figure 6.5. Finally, the model roof was lifted to
typical roofing height and left at the site until the next sampling episode.
108
Figure 6.5- Wash-off sample collection from the roof surface
6.3 Treatment and transportation of water samples
Both build-up and wash-off samples collected were labelled including information
such as the relevant intensity, duration and sample number. Additionally, deionised
water blanks and field water blanks were also included to maintain standard quality
control procedures as specified in Australia / New Zealand Standards, Water Quality
–Sampling (AS/NZS 5667.1 1998). All the samples were transported to the QUT
laboratory on the same day of sampling. The samples were tested for pH and EC and
turbidity immediately after they reached the laboratory. Sub sampling (as described
in Section 6.4 below) was done in the laboratory as early as possible and the samples
were preserved and refrigerated under a temperature of 40C as specified in Standard
Methods for the Examination of Water and Waste Water (APHA 2005) for the
analysis of physico-chemical parameters.
109
6.4 Sub sampling
Sub sampling was carried out in order to prepare representative build-up and wash-
off samples for laboratory analysis from the collected samples. Prior to sub
sampling, the weight of the build-up and wash-off samples was determined. These
weights were converted to volumes by considering the density of water. These
volumes were used to calculate the pollutant loads needed for data analysis as
discussed in Chapter 7 and Chapter 8.
The process of sub sampling was as follows:
• Firstly, the original sample was well stirred to ensure consistency of the sub
sample collected.
• Secondly, 2 L of representative total sample from each build-up and wash-off
samples was prepared.
• Thirdly, the 2 L samples were divided into 1 L containers. One 1 L sample was
used to measure pH, EC, turbidity and particle size distribution. The remaining 1
L sample was used for measuring the other physico-chemical parameters listed in
Section 6.1.
Several researchers have noted that solids can be used as an indicator of other
pollutants (Gonnetilleke et al. 2009; Mallin et al. 2008; Zafra et al. 2008).
Furthermore, the amount of pollutants attached to solids significantly varies with the
particle size of solids (Badin et al. 2008; Herngren et al. 2005; Vaze and Chiew
2002). Consequently, analysis of pollutants in different particle size fractions of
solids was important.
Therefore, it was needed to separate build-up samples into different particle size
fractions in order to investigate pollutant build-up characteristics on the study
surfaces. For this purpose, a 3 L of representative sample was taken from the original
sample and separated into four particle size ranges by wet sieving. Then the
remaining solids in each sieve were washed off into a container using deionised
water and the liquid was diluted to 1 L for the physico-chemical analysis.
110
This analysis was not undertaken for wash-off samples as the quantity of solids
collected was not considered sufficient for separation into different particle size
fractions for individual analysis. The particle size ranges which were selected were,
>300 µm, 150-300 µm, 75-150 µm, 1-75 µm and <1 µm. These particle size
fractions were selected based on past research studies of a similar nature (Herngren
et al. 2005; Sartor and Boyd 1972). The size fraction <1 µm was considered as the
potential soluble fraction in the build-up samples. This fraction was determined by
filtering a portion of total samples through 1 µm glass fiber filter paper.
Consequently, the following set of samples were subjected to physico-chemical
analysis of parameters:
• Total build-up samples;
• Wet sieved build-up samples;
• Total wash-off samples; and
• Filtrates of total build-up and wash-off samples.
6.5 Laboratory testing
Laboratory testing was conducted to measure the physico-chemical parameters
described in Section 6.1 for both build-up and wash-off samples. All the testing was
conducted using the methods specified in the standard methods (APHA 2005; US
EPA 1983; US EPA 1993). Standard laboratory procedure was followed for all the
testing. Consequently, in order to ensure the accuracy of the test data, standard
quality control procedures were followed according to methods specified in
Australia/ New Zealand Standards, Water Quality –Sampling (AS/NZS 5667.1:
1998). Additionally, for quality control purposes laboratory blanks, field blanks and
solutions of known concentration of the each analyte were included.
The following discussion provides a brief introduction to the parameters which were
measured. The complete set of test results obtained for the physico-chemical
parameters measured in build-up samples and wash-off samples are given in
Appendix B.
111
6.5.1 Particle size distribution
Malvern Mastersizer S instrument was used to determine the particle size
distribution of suspended solids. Malvern Mastersizer S instrument consists of a
sample dispersion unit which is connected by two flow cells to the optical unit
(Figure 6.6). This instrument works uses a laser diffraction technique in the analysis
of particle size distribution. A laser beam is used to create the scatter pattern from a
field of particles.
Figure 6.6- The Malvern Mastersizer S instrument
The Malvern Mastersizer S instrument uses specialised software supplied by the
manufacturer to analysis the results which are obtained from the optical unit. The
size of particles that created the scatter pattern is determined by a reverse fourier
lens. The reverse fourier lens is capable of analysing particles in the range of 0.05-
900 µm. The specified reading accuracy of the process for this range is 2% of the
volume median diameter. The interpretation of results is volume based. In this
context, the instrument analyses the volume of the particle initially and the particle
size is then determined by equating that volume to an equivalent sphere (Malvern-
Instrument- Ltd 1997).
112
Prior to testing the stormwater sample, a deionised water sample was used as the
laboratory blank to obtain the background measurement. Sample containers were
shaken gently to ensure proper mixing prior to inserting the sample into the machine.
Then the sample analysis was carried out by measuring the scatter pattern of each
sample and comparing it to the background profile generated by the blank.
6.5.2 pH, EC and turbidity
pH, EC and turbidity were measured immediately after the samples reached the
laboratory. These parameters were measured only in wash-off samples. The pH and
EC of each sample was measured by using the combined pH/ EC meter. This
instrument was pre calibrated using standard buffer solution and standard salinity
solution prior to use. The test methods used were 4500H and 2520B in the Standard
Methods for the Water and Waste Water (APHA 2005). Turbidity was measured
using a turbidity meter according to the Method 2130B in APHA (2005). The
instrument operates on the principle that light passing through a substance is
scattered by particulate matter suspended in the substance.
6.5.3 Total suspended solids and total dissolved solids
As discussed in Sections 6.2.1 and 6.2.3, the build-up samples were also collected
into a water filtration system where the samples were retained in a water column.
Consequently, Total Suspended Solids (TSS) was measured in all the total build-up
and wash-off samples and wet sieved build-up samples. This was done by filtering a
250 mL volume of sample through a 1µm glass fibre filter paper and measuring the
weight of the residue retained on it. The filter papers used were pre-washed by using
deionised water, oven dried at a temperature of 1030C-1050C and weighed before
use.
Total Dissolved Solids (TDS) was analysed by measuring the dry weight of solids
dissolved in a known volume of water. A 50 mL volume of filtrate was poured into a
pre-washed, oven dried and pre-weighed petri dish. This volume was selected to
113
ensure a noticeable increase in weight of the petri dish. Dry weight of the filtrate was
measured by determining the weight of oven dried petri dish. The test methods used
were 2540C and 2540D in the Standards Methods for the Water and Waste Water
(APHA 2005).
6.5.4 Total organic carbon and dissolved organic carbon
TOC represents the organic carbon content in total build-up and wash-off samples.
DOC represents the organic carbon content in the filtrates. Samples were prepared
according to Method 5310C (APHA 2005) for the measurement of TOC and DOC.
Shimadzu TOC-VCSH Total Organic Carbon Analyzer (Figure 6.7) was used to
measure the TOC and DOC in all the samples. The instrument is programmed with
manufacturer designed special software. It includes a blank-check program to
automatically conduct the blank check by creating and analysing ultra pure water
inside the system. TOC-VCSH unit measures TOC using an automatic sample
injection system for a wide range from 4 µg/L to 25,000 mg/L. High-concentration
samples are analysed by diluting to 25,000 mg/L with the built-in automatic dilution
function.
Figure 6.7- Shimadzu TOC-VCSH Total Organic Carbon Analyzer
114
6.5.5 Nitrogen and phosphorus parameters
The test methods used for testing nitrogen and phosphorus parameters described in
Section 6.1 are given in Table 6.2. As shown in Table 6.2, SmartChem 140 Discrete
Analyser, Seal Discrete Analyser and HACH Spectrophotometer were the main
instruments used. Both SmartChem 140 Discrete Analyser and Seal Discrete
Analyser are similar in function and capable of performing similar kind of testing.
Due to technical problems which arose with SmartChem 140 Discrete Analyser
during the research project, Seal Discrete Analyser was used for testing of some of
the nitrogen and phosphorus compounds. Additionally, a block digestion system was
used for digestion of samples for TKN and TP testing as indicated in the test
procedures.
Table 6.2- Details of the test methods used for nitrogen and phosphorus compounds Parameter Test Method Instrument
Nitrite nitrogen (NO2-)
4500 -NO2- -B (APHA 2005) SmartChem 140
Discreet Analyser
Nitrate nitrogen (NO3-)
4500 –NO3- -F (APHA 2005) SEAL Discrete
Analyser
Total kjeldahl nitrogen (TKN)
351.2 (US EPA 1993) SEAL Discrete Analyser Block digester
Total nitrogen (TN) Addition of NO2
-, NO3- and
TKN values -
Ortho-Phosphate (PO43-)
4500-P-F (APHA 2005) HACH
Spectrophotometer
Total phosphorus (TP)
365.4 (US EPA 1983) SEAL Discrete Analyser Block digestion system
Following is a brief description of each instrument listed in Table 6.2. A) Seal and SmartChem 140 Discrete Analysers
As shown in Table 6.2, both Seal and SmartChem 140 Discrete Analysers (Figure
6.8a, 6.8b) were used for testing of nitrogen and phosphorus parameters. These
115
instruments are computer controlled multi-chemistry Discrete Analysers based on
colorimetric methods. The system in each instrument is composed of a compact
benchtop chemistry unit, an external computer and an external laser printer. All
functions of the instruments are controlled by the computer and data handling, type
functions such as scheduling, reporting and quality control are permitted without
halting the instrument.
Main components of each instrument include a reagent compartment, sampling
station, sampling tray, reaction ring, aspirator and photometer. Both of these
instruments are capable of working on a number of test methods in a single run.
Detailed description of these instruments and its functionality are well documented
in instrument manuals (Inc; AQ2 Discrete Analyser operator manual 2006; Westco
Scientific Instruments).
Figure 6.8a- Seal Discrete Analyser Figure 6.8b- SmartChem 140
B) HACH DR/4000 spectrophotometer
The DR/4000 Spectrophotometer (Figure 6.9) is a direct reading instrument which is
programmed with calibrations for many tests. This instrument is mainly available in
two models. The Model DR/4000V is used for visible wavelengths and the Model
DR/4000U is used for testing in both ultraviolet and visible wavelengths. User-
entered calibrations can also be stored in the instrument. The DR/4000
Spectrophotometer provides digital readouts in direct concentration units,
absorbance or percent transmittance. When a user -generated or HACH programmed
method is selected, the on-screen menus and prompts direct the user through the
116
selected test method according to the parameter going to be tested (HACH DR 4000
instrument user manual 1999).
This instrument equipped with two modules namely single cell module and Carousel
Module. The single cell module included a 16 mm test tube adapter. The Carousel
Module includes a four place, one-inch cell carousel for one-inch square cuvettes.
The Carousel Module was selected for the testing of phosphates as specified in the
test method. More details about the instrument can be found in the DR/4000
spectrophotometer instrument manual (DR/4000 spectrophotometer instrument
manual 1999).
Figure 6.9- DR 4000 Spectrophotometer (Adapted from HACH DR 4000 instrument user manual)
C) Block digester apparatus
The block digester for preparing acid digested samples for TKN and TP testing as
stated in the test methods shown in Table 6.2. The AIM600 block digester was used
(Figure 6.10). The main components of the instrument are digestion block,
programmable controller, a set of digestion tubes, tube rack and cooling stand. The
digester consists of 50 wells to place 100 mL tubes. The programmable controller
attached to the instrument is capable of controlling digestion conditions according to
117
the methods specified. More details on this digestion system can be found in the
instrument manual (AIM600 block digestion system -user manual).
Figure 6.10- Block digester
6.6 Summary
Both build-up and wash-off samples were collected from the selected road and roof
surfaces. Build-up samples were collected from road surfaces by using specially
modified vacuum cleaner. A soft brush was used to collect build-up samples from
roof surfaces. Wash-off samples were collected by using simulating a range of
rainfall intensities for different durations for both road and roof surfaces.
Samples collected were tested based on prescribed laboratory test procedures. All the
samples were tested for a range of physico-chemical parameters, namely, pH, EC,
turbidity, particle size distribution, TSS, TDS, TOC, DOC, total and dissolved
nitrogen and phosphorus parameters.
118
119
Chapter 7 - Analysis of Pollutant Build-up
7.1 Background
As discussed in Chapter 2, pollutant build-up is the accumulation of pollutants on the
surface during dry periods and wash-off is the removal of the pollutants by rainfall
and runoff. The loads and concentrations of pollutant wash-off from urban
impervious surfaces are influenced by the amount of build-up and the composition of
the build-up pollutants (Egodawatta 2007; Pitt et al. 2004; Vaze and Chiew 2002).
Therefore, understanding the characteristics of build-up pollutants is important prior
to the analysis of pollutants wash-off.
Analysis of build-up pollutants was carried out in order to understand the nature of
the build-up on road and roof surfaces. Firstly, analysis was done for road surfaces
and roof surfaces separately due to the differences in the amount and characteristics
of pollutants which are accumulated on these surfaces (Egodawatta 2007; Furumai et
al. 2001). Secondly, pollutant build-up characteristics on both road surfaces and roof
surfaces were compared in order to understand the extent of discrimination in the
amounts of pollutants on each surface. In this context, PROMETHEE and GAIA,
one of the most common ranking techniques in multivariate data analysis was used.
7.2 Characteristics of build-up pollutants on road surfaces
As discussed in Chapter 2, quality of urban runoff is directly affected by the
characteristics of the pollutants that are accumulated on impervious surfaces (Ball et
al. 1998; Deletic and Orr 2005; Rahmat 2005). In this context, the pollutant build-up
on two road surfaces was investigated. The road sites were Drumbeat Street and Ceil
Circuit. Details of these road surfaces are provided in Chapter 5.
120
Pollutant build-up on road surfaces is typically expressed as total solids loads
(Deletic and Orr 2005; Egodawatta 2007; Sartor and Boyd 1972). Therefore, firstly,
the solids loads at each site was analysed. Secondly, the particle size distribution
analysis was carried out in order to understand the gradation of build-up solids at
each site. Finally, the physico-chemical analysis of build-up samples at each site was
carried out.
7.2.1 Analysis of total solids load
Total solids (TS) load in each sample was calculated by taking the sum of total
suspended solids (TSS) and total dissolved solids (TDS). Finally, in order to
standardize the TS load, it was converted to a load per unit area of road surface by
dividing by 3 m2 which is the plot area for build-up collection. Table 7.1 shows the
amount of total solids load at each study site investigated.
Table 7.1-Amount of total solids at each study site
Site ID Total solids load
(mg/m2)
Number of antecedent
dry days
Drumbeat Street 2595.15 14
Ceil Circuit 961.47 7
According to Table 7.1, it is evident that the solids load obtained are typical to road
sites in the Gold Coast region. According to Herngren et al. (2006) and Egodawatta
(2007), available total solids load at road surfaces in Gold Coast region is in the
range of 800-5300 mg/m2. Furthermore, there is a significant difference between
build-up loads collected from the two sites. This could be attributed to the difference
in the number of antecedent dry days. Furthermore, site specific characteristics such
as urban form, traffic volume and road surface conditions which were not
investigated in this study could be another reason for this difference. Researchers
121
have noted that build-up loads are site specific due to a diversity of factors (for
example, Egodawatta et al. 2007; Herngren et al. 2006; Vaze and Chiew 2002).
7.2.2 Particle size distribution
Particle size distribution is an important measure of the size of the solids available on
road surfaces. Particle size distribution can vary with a range of factors such as wind
and vehicular traffic and road surface conditions (Bian and Zhu 2008; Chen and
Barry 2006; Zafra et al. 2008). The composition of build-up pollutants was
determined by analysing particle size distribution. Figure 7.1 shows the cumulative
particle size distribution curves obtained for the build-up samples.
Cumulative particle size distribution at each site
0
10
20
30
40
50
60
70
80
90
100
0.1 1 10 100 1000
Particle size (µm)
Cum
ulat
ive
perc
enta
ge (
%)
DrumbeatStreet
Ceil Circuit
Figure 7.1- Variation of particle size distribution at each site
As evident in Figure 7.1, the variation of particle size distributions for the two road
surfaces is not consistent. However, such variations are common in road surface
build-up pollutants due to possible differences in antecedent dry days, traffic
volume, surface texture and surrounding landuse (Ball et al. 1998; Herngren 2005;
Zafra et al. 2008). At Drumbeat Street site more than 92% of solids are finer than
122
150 µm and at Ceil Circuit only around 77% of solids are finer than 150 µm. The
results show that the solids build-up at both road surfaces has a significant fraction of
fine particles (Particles smaller than 150 µm).
These results are in close agreement with the findings of Egodawatta (2007) and
Walker and Wong (1999) who noted a significant fraction of fine solids in pollutant
build-up in residential road surfaces in Australia. The cutoff values which they used
to define the fine solids were in the range of 125-200 µm. Egodawatta (2007) who
investigated a set of residential landuses in Gold Coast, Australia found that more
than 70% of solids are available in the particle size fractions which are finer than 200
µm. This is further supported by the findings of Walker and Wong (1999) who noted
that up to 70% of the solids are finer than 125 µm in road deposited solids in a
number of road surfaces in Australia.
7.2.3 Physico-chemical characteristics of build-up pollutants
As discussed in Chapter 2, quality of urban runoff is directly affected by the
characteristics of the pollutants that are accumulated on road surfaces (Deletic and
Orr 2005; Rahmat 2005). In this context, investigation of physico-chemical
characteristics of pollutant build-up parameters is important. Therefore, total build-
up samples which were collected from the road surfaces were analysed for a range of
physico-chemical parameters as described in Chapter 6.
The nitrogen and phosphorous compounds which were analysed include nitrite
nitrogen (NO2-), nitrate nitrogen (NO3
-), total kjeldahl nitrogen (TKN), total nitrogen
(TN), orthophosphate (PO43-) and total phosphorus (TP). TN was obtained by taking
the sum of NO2-, NO3
- and TKN. Additionally, Total organic carbon (TOC) was also
included in the analysis. In order to standardise the data derived, the amount of
pollutants measured in each build-up sample at each study site were converted to
load per unit area of the road surface. The results are presented in Table 7.2. The
original test results are presented in Appendix B.
123
Table 7.2- Total pollutants loads at each study site (mg/m2)
Site ID
TOC
NO2-
NO3-
TKN
TN
PO43-
TP
Drumbeat
Street 72.92 0.03 1.90 21.21 23.14 7.09 7.80
Ceil
Circuit 19.34 0.00 0.82 8.54 9.35 1.58 1.82
In comparison to past several research studies, the pollutant loads found in this study
sites are relatively low. For example, Sartor and Boyd (1972) found that the surface
loading for TN varied between 14.3 and 51.8 g/m2 and between 2.95 and 8.43 g/m2
for TP. However, the loadings of these pollutants show inherent variability among
urban road surfaces. The factors influencing this variability include road surface
condition, number of antecedent dry days, nature of anthropogenic activities,
climatic condition and catchment management practices (Barrios 2000; Hope et al.
2004; Sartor and Boyd 1972). The lower pollutant loads collected from these sites
could be attributed to influence of these factors.
Furthermore, the pollutant loads are significantly different between the two sites.
Drumbeat Street site shows higher loads for all the pollutant species in comparison
to the Ceil Circuit site. This is primarily due to the higher build-up load collected
from Drumbeat Street. Furthermore, total organic carbon content in Drumbeat Street
site is significantly higher than the total organic carbon content in Ceil Circuit. The
reason for this difference could be attributed to the higher amount of vegetation at
the Drumbeat Street site in comparison to the Ceil Circuit site. This could also be a
reason for the increased amount of nutrients at Drumbeat Street site (Flanagan and
Forster 1989). Flanagan and Forster (1989) suggested that the disproportionately
larger surface area of organic matter can hold nutrients and this will cause an
increase in the nutrients load.
124
The weight of each pollutant per unit weight of total solids in build-up was derived
as listed in Table 7.3. As evident in Table 7.3 a considerable weight of the total
solids is attributed to organic carbon. This agrees well with the findings of Roger et
al. (1998) who noted a high organic carbon content in solids build-up.
Table 7.3- Amounts of pollutants per unit weight of total solids (mg/g)
Site ID
TOC
NO2-
NO3-
TKN
TN
PO43-
TP
Drumbeat
Street 28.10 0.01 0.73 8.17 8.92 2.73 3.00
Ceil
Circuit 20.11 0.00 0.85 8.88 9.73 1.64 1.89
7.2.4 Investigation of pollutants in different particle size fractions of solids
It is well understood that the amount of pollutants in build-up significantly vary with
the particle size of the solids (Egodawatta 2007; Herngren et al. 2006; Vaze and
Chiew 2002). Therefore, pollutant types in different particle size fractions of solids
were separately tested as discussed in Chapter 6. The size ranges used were >300
µm, 300-150 µm, 150-75 µm, 75-1 µm and <1 µm. Particle size fraction <1 µm was
considered as the potential soluble fraction of the build-up.
The wet sieved build-up samples were analysed for the same physico-chemical
parameters as above. Based on the laboratory test results, the pollutant load in each
particle size fraction was obtained per unit weight of solids build-up by considering
the total solids load at each site. The results are shown in Table 1 (Appendix C) and
the graphical representation of the amount of pollutants in each particle size fraction
is shown in Figure 7.2a, 7.2b.
125
Figure 7.2a- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for Drumbeat Street
Figure 7.2b- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for a) Ceil Circuit It was noted that a higher fraction of solids are finer than <150 µm. This further
supports the findings of other researchers (Ball et al. 1998; Egodawatta 2007;
Walker and Wong 1999) who concluded that road surfaces in Australia have a
relatively high proportion of fine solids as discussed in Section 7.2.2. Furthermore, a
relatively higher amount of nitrogen and phosphorus compounds are in the particle
size range <150 µm. This confirms the highly polluted nature of the finer fraction of
126
pollutant build-up (Bian and Zhu 2008; Vaze and Chiew 2002). Vaze and Chiew
(2002) in their study found that more than 60% of TN and TP were attached to
particles below 150 µm.
Most of the particle size fractions contain significantly higher loads of TKN which is
the organic form of nitrogen. This indicates that TKN is the dominant form of
nitrogen compound in pollutant build-up. This is further supported by the increase in
the amount of TN with the increase in TOC in each particle size fraction as seen in
Figure 7.2a, 7.2b.
7.3 Characteristics of build-up pollutants on roof surfaces
Although, road surfaces are among the most critical contributors to urban stormwater
pollution, the role of other types of impervious surfaces on urban stormwater runoff
should also be investigated. In this regard, roofs surfaces are important as it has been
identified as an important contributor to urban stormwater pollution (Bannerman et
al. 1993; Egodawatta 2007). Chang and Crowley (1993) noted that roofs may have a
significant influence on stormwater quality as they may make up more than 50% of
the impervious surfaces in residential areas. According to Forster (1999), roofs can
play a major role in the pathway that pollutants travel between the atmosphere to
receiving water bodies, because roofs are efficient collectors of particles fallout from
the atmosphere and efficient deliverers of these particles to urban stormwater runoff.
However, only a limited number of research studies have been undertaken on
pollutant build-up on roof surfaces (Egodawatta 2007; Van Metre and Mahler 2003).
Egodawatta (2007) carried out a detailed investigation of the characteristics of
pollutant build-up on roofs by using two model roof surfaces consisting of two
different roofing materials (corrugated steel and concrete tiles). From his study it
was noted that pollutant build-up was independent of the roofing material. Therefore,
the difference in roofing material was not considered in the data analysis undertaken
in this research study.
127
As discussed in Chapter 6, build-up samples on roof surfaces were collected in three
sampling episodes. BU1, BU2 and BU3 represent the average amount of pollutants
at both roof surfaces for different sampling episodes (See Table 2 in Appendix C).
The antecedent dry days for the collection of samples BU1, BU2 and BU3 were 8, 6
and 6 respectively. Similar to the build-up analysis for road surfaces, firstly the total
solids load for each build-up sample was analysed. Secondly, particle size
distribution analysis was carried out. Finally, the physico-chemical analysis of build-
up was carried out in order to understand the nature of the roof surface build-up.
7.3.1 Analysis of total solids load
Similar to the road surfaces, the amounts of total solids (TS) present was calculated
by taking the sum of total suspended solids (TSS) and total dissolved solids (TDS)
measured in each sample. Finally, in order to standardise the results, the amount of
total solids was converted to load per unit area of the roof surface as shown in Table
7.4.
Table 7.4-Average total solids load (mg/m2)
Sample ID Total solids load
BU 1 190
BU 2 190
BU 3 180
It was noted that the build-up loads collected from the three sampling episodes are in
the same order. The build-up loads are typical amounts recovered from roof surfaces
in past research studies (Furumai et al. 2001; Van Metre and Mahler 2003). Van
Metre and Mahler (2003) found that build-up on roof surfaces can vary in the range
of 160 to 1200 mg/m2 depending on the magnitude of the antecedent dry period.
However, according to Egodawatta (2007) who carried out investigations on the
same roof surfaces, at Gold Coast, the pollutant build-up for 7 days of antecedent dry
period was around 800 mg/m2. This is considerably higher than the build-up load
128
found in this research study. This is attributed to factors such as the nature of
anthropogenic activities in the surrounding area and climatic conditions (Gromaire-
Mertz et al. 1999; Van Metre and Mahler 2003). This again highlights the highly
variable nature of pollutant build-up and the significant influence exerted by various
factors.
In comparison to road surfaces, the solids load found on roof surfaces are relatively
low. This could be attributed to the difference in surface characteristics such as
surface roughness and slope and different pollutant sources. Furthermore, the
contribution of solids from different sources such as vehicular activities has a lesser
impact on roof surfaces compared to road surfaces.
7.3.2 Particle size distribution
Similar to the analysis of road surfaces, gradation of build-up pollutants was
determined by analysing particle size distribution. Figure 7.3 shows the cumulative
particle size distribution curves obtained for each build-up sample.
Cumulative particle size distribution at each Build-up sample
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
0.1 1 10 100 1000
Particle size (µm)
Cum
ulat
ive
perc
enta
ge (
%)
BU 1
BU 2
BU 3
AVG ofBU1,BU2 andBU3
Figure 7.3- Cumulative particle size distribution of each build-up sample
129
As evident in Figure 7.3, around 80% of the solids are finer than 150 µm for all the
build-up samples. Consequently, build-up on roof surfaces also contains a significant
amount of fine particles similar to the road surfaces. This can be further supported by
the study of Egodawatta (2007) who noted that around 72% of solids are finer than
200 µm after investigating roof surface build-up for 7 days of antecedent dry period
in the Gold Coast region.
However, the percentage of finer fraction of solids noted in roof surfaces (which is
over 85%) is much higher compared to the percentage of finer fraction of solids in
Ceil Circuit site (which is around 70%). This could be attributed to the fineness of
atmospheric depositions. Furthermore, due to the reduced texture depth and greater
slope of roof surfaces, larger particles may not remain on roof surfaces. This agrees
well with the findings of Egodawatta (2007) who noted significantly fine solids in
roof surfaces compared to the road surfaces.
As shown in Figure 7.3, particle size distribution curves for all three sampling
episodes are quite similar. This is very different to the results obtained for the road
surfaces where two distinct particles size distribution profiles were noted. This
would suggest that only limited re-distribution occurs on roof surfaces and the
initially deposited materials remain on the surface for a relatively longer period of
time when compared to the road surfaces. As noted by Egodawatta (2007), this could
be attributed to the reduced influence of vehicular induced wind turbulence.
Furthermore, the relatively similar particle size distribution profiles on roof surfaces
could also be attributed to relatively low texture depth on roof surfaces.
7.3.3 Physico-chemical characteristics of build-up pollutants
Similar to the road surfaces, using the data obtained from laboratory testing, the total
pollutant load for each build-up sample was calculated per unit area of the roof
surface as given in Table 7.5.
130
Table 7.5- Pollutants loads in each build-up sample (mg/m2)
Sample
Name
TOC
NO2-
NO3-
TKN
TN
PO43-
TP
BU1 2.12 0.31 0.64 2.66 3.61 5.78 6.25
BU2 8.38 0.50 0.86 4.10 5.46 7.23 7.80
BU3 5.22 0.30 0.58 1.16 2.05 0.97 1.65
As shown in Table 7.5, pollutant loads exhibit significant variation among the three
sampling episodes. This can be attributed to the activities in the vicinity of the roof
surfaces before the each sampling episode. For example, higher loads of pollutants in
BU2 could be due to the lawn mowing operation which occurred on the day of the
sample collection.
The ratio of NO2- to NO3
- which is around 0.5 for each sampling episodes is
relatively high compared to the ratio for road surfaces which was negligible. NO2- is
relatively unstable and is oxidised to NO3- readily (Chapmen 1992). However,
compared to the road surfaces, roof surfaces are subjected to higher direct
atmospheric deposition which produces higher NO2- for the roof surface build-up.
This is further confirmed by the higher NO2- to NO3
- ratio in BU2 sample which was
subjected to unusual atmospheric depositions as noted previously.
In comparison to the pollutant loadings for 7 days at the road surfaces, the nutrients
load per unit area is considerably low for the roof surfaces. This is due to the low
build-up loads at roof surfaces compared to road surfaces. Furthermore, this could
also be attributed to different sources of nutrients at each surface. For example,
vehicular traffic, soil erosion may increase the nutrient loading on road surfaces
(Novotny and Chesters 1981).
131
The weight of each pollutant per unit weight of total solids was obtained. As evident
in Table 7.6, unlike road surfaces, similar to the contribution of organic carbon load,
total nitrogen and total phosphorus also contribute considerable loads to total solids
load at roof surfaces. In the case of the road surfaces, the higher amount of organic
matter could be generated from direct influence of vehicle exhaust, vegetation debris
and soil particles which would increase the organic carbon content in solids build-up
on road surfaces (Rogge et al. 1993).
Table 7.6- Amounts of pollutants per unit weight of total solids (mg/g)
Sample
Name
TOC
NO2-
NO3-
TKN
TN
PO43-
TP
BU1 11.16 1.62 3.38 14.02 19.02 30.43 32.87
BU2 44.11 2.63 4.55 21.56 28.73 38.05 41.03
BU3 29.00 1.68 3.24 6.46 11.38 5.39 9.16
7.3.4 Investigation of pollutants in different particle size fractions of solids
The pollutant loads in different particle size fractions of solids build-up was
investigated for roof surfaces in order to understand the characteristics of pollutants
in each particle size fraction. The results are shown in Table 3 in Appendix C and
Figure 7.4a, 7.4b, 7.4c.
132
Figure 7.4a- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU1
Figure 7.4b- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU2
133
Figure 7.4c- Graphical representation- Concentration of pollutants in different
particle size fractions of build-up for BU3
According to Table 3 (see Appendix C) and Figure 7.4a, 7.4b, 7.4c, a significant
fraction of solids are finer than 1 µm for all the collected samples. This would again
be attributed to the fineness of atmospheric deposition on roof surfaces. According to
Quek and Forster (1993) and Gromaire-Mertz et al. (1999), roof weathering and dry
deposition are the main sources of fine solids on roof surfaces. Furthermore, in
comparison to the road surfaces, all the particle size fractions contain a lower amount
of solids compared to roof surfaces. This is mainly attributed to the lower solids load
on the roof surfaces compared to road surfaces.
Particle size <1 µm which is the potential soluble fraction of the build-up contains
significantly higher amount of NO2-, NO3
- in each sample. This indicates the high
solubility of nitrogen compounds. This confirms the high solubility of NO2- and
NO3-.
The particle size fraction <150 µm contains a higher amount of TOC, nitrogen and
phosphorus compounds in comparison to the particle size fraction >150 µm. These
results are similar to the road surfaces where the particle size fraction <150 µm was
134
found to be the more polluted than the particle size fraction >150 µm. Furthermore,
as noted in the analysis of pollutant build-up on road surfaces, TN shows an
increasing trend with the increase in TOC.
7.4 Comparison of pollutant build-up characteristics on road surfaces and roof surfaces
The above analysis outlines the primary characteristics of the build-up samples
collected from road surfaces and roof surfaces. As the main focus of this research is
to identify a set of surrogate parameters for selected water quality parameters, it is
essential to compare the pollutant build-up characteristics on both road and roof
surfaces. This understanding is important to identify a common set of parameters as
surrogates for both roads and roof surfaces.
Considering the limited number of total build-up samples for both road and roof
surfaces, only the wet sieved build-up samples were subjected to PROMETHEE and
GAIA analysis in order to understand the physico-chemical behaviour of parameters.
The samples in different particle size ranges were considered as objects and a range
of physico-chemical parameters were considered as variables. The variables (criteria)
used for the PROMETHEE and GAIA analysis were total solids (TS), total organic
carbon (TOC), nitrite (N2), nitrate (N3), total kjeldahl nitrogen (TKN), total nitrogen
(TN), orthophosphates (P4) and total phosphorus (TP).
For the analysis, all the parameters were given the same weighting and preference
function. For PROMETHEE as discussed in Chapter 4, the user should specify the
preferred ranking order which is a maximise or minimise function. In this study the
variables were maximised so that the most polluted sample in terms of above
variables were ranked first in the PROMETHEE analysis. All the variables were
given the same weighting and hence no variables were favoured over the other.
Furthermore, the preference function was set to V-shape, which means that
preference threshold P, representing the smallest deviation considered decisive was
135
used in processing the data. Moreover, P was set to the maximum value of each
variable. Finally, concentration below the detection limit was set to half the detection
limit of the specific parameter (Guo et al. 2004; Herngren et al. 2005). The following
discussion is based on the outcomes of the PROMETHEE and GAIA analysis of
pollutant build-up data from both road and roof surfaces.
Table 7.7 gives PROMETHEE results and Figure 7.5 gives the GAIA biplot for all
the wet sieved build-up data for both road and roof surfaces. The pollutant loads
obtained in the form of mg/g of total solids was used in the analysis. GAIA biplots
display total variance of (∆) 84.77% which indicates that most of the information is
presented in the analysis. PROMETHEE ranked the particle size classes from worst
to best in terms of pollutant concentrations.
136
Table 7.7- PROMETHEE 2 ranking
Note: C Ceil Circuit D Drumbeat Street R1 Build-up sample in first sampling episode R2 Build-up sample in second sampling episode R3 Build-up sample in third sampling episode
Sample Net Φ Ranking order
D1-75 0.55 1
D75-150 0.39 2
C75-150 0.31 3
D<1 0.23 4
<1R2 0.14 5
C1-75 0.05 6
D150-300 0.03 7
<1R1 0.02 8
D>300 0.01 9
75-150R2 0.00 10
<1R3 -0.04 11
75-150R1 -0.05 12
C<1 -0.05 13
C150-300 -0.06 14
C>300 -0.07 15
1-75R2 -0.12 16
150-300R2 -0.12 17
>300R2 -0.13 18
75-150R3 -0.13 19
150-300R1 -0.15 20
>300R1 -0.15 21
150-300R3 -0.16 22
>300R3 -0.17 23
1-75R3 -0.17 24
1-75R1 -0.18 25
137
Figure 7.5- GAIA analysis for build-up samples Note: C Ceil Circuit D Drumbeat Street
Variables Ceil Circuit site samples
Drumbeat Street site samples First sampling episode- roof surfaces Second sampling episode- roof surfaces Third sampling episode- roof surfaces
As evident in the GAIA biplot, build-up samples from road surfaces clearly
discriminate from roof surface build-up samples. Furthermore, all the variables point
towards the samples from road surfaces indicating the highly polluted nature of road
surface build-up in terms of all the pollutants in comparison to the roof surfaces.
138
This is further confirmed by Table 7.7, where almost all the samples from road
surfaces are ranked first while samples from roof surfaces are ranked last. This
confirms that significantly different pollutant characteristics of these surfaces.
Therefore, from an analytical perspective, in further analysis roads and roofs were
considered separately.
Additionally following conclusions can be derived from Table 7.7 and Figure 7.5.
• TP is strongly correlated to PO43-. According to Table 1 and Table 3 (Appendix
C) and Figure 7.2a, 7.2b and Figure 7.4a, 7.4b, 7.4c, the high amount of TP is
attributed to PO43-. This suggests that PO4
3- is the dominant form of
phosphorus.
• TOC is strongly correlated to TN. This further confirms the conclusions made
from Table 1 and Table 3 (Appendix C) where TN increases with the increase
in TOC.
• Additionally, this analysis confirms the highly polluted nature of the particle
size fraction <150 µm in terms of solids, organic matter and nutrients. As seen
in the biplot, since the decision vector π points along the PC1 axis and point
towards the particle size fraction <150 µm samples from road surfaces, it can
be confirmed that this is the most polluted particle size fraction in terms of all
the pollutants.
• According to Figure 7.5, particle size fractions >150 µm show clusters. This
suggests that these particle sizes have similar behaviour for all the pollutants.
As these objects lie opposite to the direction of the π vector, it can be said that
this particle size fraction is the least polluted. This is further confirmed by the
PROMETHEE results. As shown Table 7.7, samples of particle size faction
>150 µm is ranked last among the other particle size fractions for both road and
roof surfaces.
139
7.5 Conclusions
Following conclusions were derived from the analysis of pollutant build-up on road
and roof surfaces;
• Pollutant build-up characteristics identified for both road surfaces and roof
surfaces are in general agreement with past research studies.
• A high percentage of solids from both roads and roof surfaces are smaller than
150 µm. This is the most polluted particle size fraction for both surface types.
• Road surfaces show significantly higher loads of pollutants compared to roof
surfaces. PROMETHEE and GAIA analysis clearly indicates the
discrimination between the road and roof surfaces.
140
141
Chapter 8 - Analysis of Pollutant Wash-off
8.1 Background
The main focus of this Chapter was the identification of a set of surrogate parameters
for the other stormwater quality parameters. For this, samples which were generated
by simulating rainfall events on road and roof surfaces were used as discussed in
Chapter 6. Therefore, prior to the analysis of surrogate parameters, a separate
analysis was done to check the appropriateness of the simulated wash-off process
undertaken in this study compared to the general knowledge on the wash-off process
on urban impervious surfaces.
In this context, concentration of total solids in collected wash-off samples for each
intensity and duration and particle size distribution of wash-off solids were analysed.
The pollutant wash-off process has been identified as a dependent process of
pollutant build-up (Duncan 1995). Therefore, wash-off samples from road surfaces
and roof surfaces were analysed separately as the build-up on these two types of
surfaces are considerably different as discussed in Chapter 7.
Analysis of physico-chemical parameters was carried out by using two common
multivariate data analysis techniques, namely, Principal Component analysis (PCA)
and Partial Least Squares (PLS). These techniques were employed to understand the
linkages among parameters and thereby to identify the surrogate water quality
parameters. Consequently, PCA was used to identify potential surrogate parameters
for parameters of interest. PLS was used to check the validity of the selected
parameters as surrogate parameters. For this purpose, a number of models were
developed for the identified surrogate parameters and key water quality parameter of
interest.
142
8.2 Understanding the solids wash-off process
8.2.1 Road surfaces
As discussed in Chapter 4 and Chapter 6, six rainfall intensities were simulated on
road surfaces, namely, 20, 40, 65, 86, 115 and 135 mm/hr. Wash-off samples
collected from these intensities were analysed for a range of physico-chemical
parameters as discussed in Chapter 6. The following data analysis was based on total
solids concentration and particle size distribution data obtained from the laboratory
analysis.
A) Variation of total solids concentration with rainfall duration
The total solids (TS) concentration of each wash-off sample was calculated by taking
the sum of total suspended solids (TSS) and total dissolved solids (TDS) which were
measured in each wash-off sample. Figure 8.1a, 8.1b show the variation of total
solids concentration with duration for all intensities. As seen in Figure 8.1a, 8.1b, the
concentration of total solids in wash-off was higher for the shorter durations for all
the intensities compared to the longer durations. The concentration decreases
exponentially with the increase in duration for all the intensities. This agrees with the
general understanding of the wash-off process as noted by researchers such as
Deletic (1998), Duncan (1995) Egodawatta (2007), and Weeks (1981).
143
0
100
200
300
400
500
600
700
0 10 20 30 40 50
Rainfall duration (min)
Was
h-of
f TS
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.1a- Variation of TS concentration with rainfall duration and intensity for Drumbeat Street
0
50
100
150
200
250
300
350
0 10 20 30 40 50Rainfall duration (min)
Was
h-of
f TS
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.1b- Variation of TS concentration with rainfall duration and intensity for Ceil Circuit
144
Solids concentration in the wash-off is relatively higher at the Drumbeat Street site
in comparison to the Ceil Circuit site. This could be attributed to the higher pollutant
load in the build-up at Drumbeat Street site as discussed in Chapter 7. This
conclusion is supported by the findings of Egodawatta (2007) who noted that the
wash-off pollutant load is significantly influenced by the initially available pollutant
loads on road surfaces. Additionally, the differences in surface conditions such as
texture depth and the surface slope could be attributed to the difference in solids
wash-off concentrations between the sites (Egodawatta 2007; Hope et al. 2004).
B) Particle Size distribution
Particle size distribution is an important characteristic of solids wash-off (Hoffman
et al. 1984; Pitt et al. 1995). It indicates the available particle size fractions of solids.
Therefore, in order to obtain a detailed understanding of solids wash-off process, the
particle size distribution of each sample was analysed.
The particle size distribution data was obtained as volumetric percentages for
different durations for the six intensities simulated. Initially, the average volumetric
percentages were obtained by considering the particle size distribution measurements
for all the durations for each intensity. Finally, the cumulative particle size
distribution curves were plotted for all six intensities for both study sites. The
variation of particle size distribution of solids in the wash-off with rainfall intensity
is shown in Figure 8.2a, 8.2b for each study site separately. Furthermore, the particle
size distribution of pollutant build-up at each study site is also shown in order to
compare the wash-off behaviour of solids with the pollutant build-up.
145
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
0.10 1.00 10.00 100.00 1000.00Particle size (µm)
Cum
ulat
ive
perc
enta
ge(%
)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Averagewash-offBuild-uppollutants
Figure 8.2a- Variation of particle size distribution with rainfall intensity for
Drumbeat Street
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
0.10 1.00 10.00 100.00 1000.00Particle size (µm)
Cum
ulat
ive
perc
enta
ge (%
)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Averagewash-offBuild-uppollutants
Figure 8.2b- Variation of particle size distribution with rainfall intensity for
Ceil circuit As seen in Figure 8.2a, 8.2b, the wash-off for intensities 115 and 135 mm/hr contain
relatively more fine particles (<150 µm) when compared to other intensities at both
sites. This suggests that the wash-off of finer fraction of solids increases with the
146
increase in the rainfall intensity. This confirms the findings of past researchers who
noted that higher intensities have greater capacity for removing finer particles from
an impervious surface than the lower intensities (for example, Chui 1997; Pitt et al.
2004). This is attributed to the relatively higher transport capacity of runoff of high
intensity rain events as noted by Vaze and Chiew (2002). However, even for high
intensity rain events, the amount of wash-off of fine particles is relatively low,
compared to the amount of fine particles available in the pollutant build-up. This
confirms findings of past researchers who noted that only a fraction of build-up is
washed off even for high intensity rain events (for example, Vaze and Chiew 2002).
As evident in Figure 8.2a, 8.2b, the percentage of wash-off of fine particles during
relatively higher intensities (86, 115 and 135 mm/hr) show significantly higher
values for Drumbeat Street in comparison with Ceil Circuit site. Analysis of
pollutant build-up in each study site, revealed that the solids load in the build-up at
Drumbeat Street site contain a higher fraction of finer particles compared to the Ceil
Circuit site. This confirms that the pollutant wash-off process is always dependent on
the pollutant build-up. Consequently, it can be surmised that the wash-off process
investigated in this study is in agreement with the general understanding of the wash-
off process on road surfaces.
8.2.2 Roof surfaces
Similar to the road surfaces, six intensities were simulated on the two roofs surfaces
during three sampling episodes as follows:
• 65, 86 mm/hr intensities in the first sampling episode;
• 115, 135 mm/hr intensities in the second sampling episode; and
• 20, 40 mm/hr intensities in the third sampling episode.
20, 86 and 135 mm/hr intensities were simulated on the steel roof surface and
remaining intensities namely, 40, 65 and 115 mm/hr were simulated on the tile roof
147
surface. Egodawatta (2007) who carried out wash-off investigations using the same
roof surfaces noted that the wash-off process was independent of the type of roofing
material. Therefore, the type of roofing material was not considered as a variable in
the data analysis. Similar to the road surfaces, wash-off behaviour of solids was used
as the indicator to understand the wash-off process on roof surfaces.
A) Variation of total solids concentration with rainfall duration
Total solids (TS) concentration of each wash-off sample was calculated by taking the
sum of total suspended solids (TSS) and total dissolved solids (TDS) as discussed in
Chapter 6. Variation of total solids concentration with rainfall duration for all the
intensities was analysed. The results are shown in Figure 8.3. Similar to the road
surfaces, the TS concentration decreases exponentially with duration for all the
intensities. This is again in general agreement with total solids wash-off behaviour
on roof surfaces (Forster 1999; Quek and Forster 1993; Van Metre and Mahler
2003).
0.0
50.0
100.0
150.0
200.0
250.0
0 2 4 6 8 10 12 14
Rainfall duration (min)
Was
h-of
f TS
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.3- Variation of TS concentration with rainfall duration and intensity
for roof surfaces
148
However, in comparison to the road surfaces, the roof surfaces had relatively low
total solids concentrations. This is attributed to the low solids build-up on roof
surfaces compared to road surfaces. Additionally, total solids concentration of wash-
off samples from roof surfaces decreases much more rapidly for longer durations
compared to the solids concentration for longer durations for road surfaces (see
Figure 8.1 and Figure 8.3). This can be attributed to the relatively faster roof runoff
due to the steeper slope and relatively smooth surface of roof surfaces compared to
road surfaces (Berdahl et al. 2008; Furumai et al. 2001).
B) Particle Size distribution
Similar to the road surfaces, particle size distribution analysis of wash-off samples
was carried out in order to understand the gradation of solids in wash-off from roof
surfaces. Consequently, cumulative particle size distribution curves were plotted for
all six intensities. The variation of particle size distribution with rainfall intensity is
shown in Figure 8.4. Furthermore, the particle size distribution curve for the average
of the three build-up sampling episodes also given in Figure 8.4 for comparison
purposes.
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
0.1 1 10 100 1000
Particle Size (µm)
Cum
ulat
ive
perc
enta
ge (%
)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Averagewash-offBuild-upaverage
Figure 8.4- Variation of particle size distribution with rainfall intensity for roof
surfaces
149
According to Figure 8.4, it is evident that on average, around 50% of particles are
finer than 150 µm in the wash-off from roof surfaces. This suggests that there is a
considerable contribution of fine particles from the roof surfaces to stormwater
runoff. In comparison to the road surfaces, the percentage of fine particles (particle
size fraction <150 µm) in the wash-off from roof surfaces is higher for the majority
of the rainfall intensities. Therefore, roof surfaces can have a major influence on
stormwater runoff quality as the finer particles have been proven to be more polluted
than the coarser particles.
Furthermore, as seen in Figure 8.4, for each sampling episode, the percentage of fine
particles in the wash-off increases with the increase in rainfall intensity. For
example, for the 40 mm/hr intensity around 80% of the particles were less than 150
µm and for the 20 mm/hr intensity the wash-off was around 50%. This suggests that,
higher intensities are more capable of removing pollutants from roof surfaces than
the lower rainfall intensities as noted by Yaziz et al. (1989). Consequently, it can be
said that, these findings agree well with the findings of past researchers who
investigated the wash-off behaviour of solids on roof surfaces (Egodawatta 2007;
Furumai et al. 2001).
8.2.3 Comparison of pollutants concentrations on road and roof surfaces
As noted in Chapter 7, pollutant build-up on road and roof surfaces are significantly
different to each other. Since pollutant wash-off is dependent on the amount of
pollutant build-up on the surface, it was important to understand the variability of
pollutant concentrations in the wash-off from road and roof surfaces. For this
purpose, firstly, pollutant concentrations in wash-off from roads and roofs were
compared. Table 8.1, shows the mean and standard deviation of measured
parameters for both surfaces.
For all the parameters except phosphorus, roof surfaces show significantly lesser
mean values of pollutant concentrations compared to road surfaces. This indicates
that pollutant concentrations of roof runoff are relatively low compared to the
150
pollutant concentrations from road surfaces. The results confirmed the findings of
several researchers who noted that pollutants concentrations in runoff from roof
surfaces are significantly lower compared to runoff from road surfaces (Chebbo and
Gromaire 2004; Furumai et al. 2001; Gnecco et al. 2005; Gromaire-Mertz et al.
1999; Huang et al. 2007). Chebbo and Gromaire (2004), Gromaire-Mertz et al.
(1999) and Pazwash and Boswell (1997) noted that there is very low suspended
solids concentrations from roof runoff compared to runoff from road surfaces. For
example, Gromaire-Mertz et al. (1999) found that the event mean concentration of
suspended solids in roof runoff was in the range of 3-304 mg/L whilst suspended
solids concentration in road runoff varied from 49-498 mg/L.
According to Huang et al. (2007) concentration of total nitrogen in roof runoff was
around 2.57 mg/L whilst total nitrogen concentration in road runoff was around 3.58
mg/L. On the other hand, standard deviation of pollutant concentrations of roof
runoff is significantly low compared to the standard deviation of pollutant
concentrations of road runoff. This indicates the low variability of pollutant
concentration of roof runoff. As the pollutant build-up load on roof surfaces is
significantly lower than the pollutant build-up load on road surfaces and as roof
runoff is faster compared to the road runoff, pollutants are easily washed off from
the roof surfaces within a short duration. Consequently, the concentration of
pollutants in roof runoff could be low due to the limited pollutant availability for
wash-off with the runoff as a storm progresses. Therefore, roof surfaces can be
defined as a source limiting surface while road surfaces as a transport limiting
surface.
Additionally, in order to develop an understanding of the differences in pollutant
concentrations in the wash-off from each surface, multivariate analysis in the form of
PCA was carried out. PCA was carried out for all the wash-off samples collected
from both roads and roof surfaces .The biplot obtained is given in Figure 8.5.
151
Table 8.1- Mean concentration and standard deviation values of measures parameters
Study site Parameter EC (µS/cm)
TS (mg/L)
TTU (NTU)
TOC (mg/L)
TNO2-
(mg/L) TNO3
- (mg/L)
TKN (mg/L)
TN (mg/L)
TPO43-
(mg/L) TP
(mg/L)
Mean 69.63 265.71 26.57 24.99 0.00 0.13 2.70 2.82 0.45 0.56 Drumbeat Street
STD 29.87 184.52 22.92 30.61 0.00 0.13 2.38 2.49 0.37 0.46
Mean 57.91 130.32 7.96 13.23 0.00 0.17 2.96 3.13 0.20 0.28 Ceil Circuit
STD 21.13 62.83 6.91 9.32 0.00 0.06 2.14 2.18 0.13 0.14
Mean 60.48 65.91 3.69 3.11 0.03 0.13 0.54 0.70 1.80 1.85 Roofs
STD 13.94 49.30 3.24 1.89 0.00 0.08 0.25 0.32 1.45 1.71
Note: TTU- Turbidity; EC- Electrical conductivity; TNO2-- Total nitrite-nitrogen; DNO2
-- Dissolved nitrite-nitrogen TOC- Total organic carbon; TNO3-- Total
nitrate- nitrogen; DNO3-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; TN- Total nitrogen; TS- Total solids; TPO4
3-- Total Phosphates; TP- Total
phosphorus.
152
Figure 8.5- Biplot for all the physico-chemical parameters for both roads and
roof surfaces Note: TTU- Turbidity; EC- Electrical conductivity; TNO2- Total nitrite-nitrogen; TOC- Total organic
carbon; TNO3- Total nitrate- nitrogen; TN- Total nitrogen; TKN- Total kjeldahl nitrogen; TS- Total
solids; TP- Total phosphorus; TPO4- Total Phosphates.
As shown in Figure 8.5, samples from road surfaces clearly discriminate from the
samples from roof surfaces. According to Figure 8.5, majority of the road surface
samples show positive scores on PC2 whereas almost all the roof surface samples
show negative scores on PC2. On the other hand, phosphorus shows higher negative
loading on PC2 whereas TN, TOC, TTU show higher positive loadings on PC2. This
indicates that, these surfaces are significantly different in wash-off characteristics.
Therefore, from an analytical perspective, separate analysis of roads and roofs was
undertaken.
153
8.3 Analysis of physico-chemical parameters
The main focus of this research study was to identify a set of easy to measure
surrogate parameters to determine pollutant concentrations in urban stormwater
runoff. Therefore, data analysis of physico-chemical parameters was carried out with
the aim of identifying important correlations among the physico-chemical water
quality parameters. Analysis was carried out separately for road surfaces and roof
surfaces.
Data analysis was carried out using two common multivariate data analysis
techniques; Principal Component Analysis (PCA) and Partial Least Squares (PLS).
A detailed discussion of the application of PCA and PLS is available in Chapter 4.
Prior to the application of PCA and PLS, data was pre-treated to remove the
skewness which would affect the statistical significance of results (Ayoko et al.
2007; Goonetilleke et al. 2005; Kokot et al. 1998; Zhang et al. 2006). The final data
matrices were then subjected to PCA and PLS analysis for pattern recognition and
for identification of linkage between selected parameters with other physico-
chemical parameters. The following discussion provides a detailed description of
data preprocessing and application of PCA. Application of PLS is discussed in
Section 8.5.
Data preprocessing
Prior to multivariate data analysis, concentrations below detection limit were set to
half of the detection limit of each parameter. Then the values from different sites
were rectified to eliminate any bias due to different build-up loads at different sites
as the investigation of pollutant build-up revealed significant differences in pollutant
loads at each study site. Therefore, data was standardized based on the build-up load
at each site. Hence, the values used for further analysis was in the form of mg/L/g of
build-up load (See Table 1, Table 2 and Table 3 in Appendix D).
154
The data from both road and roof surface were arranged into two separate data
matrices. The matrix containing road surfaces data consisted of 68 objects and 19
variables. The matrix containing roof surfaces data consisted of 34 objects and 19
variables. The columns in the matrices represented measured water quality
parameters and rows represented samples for each duration for each rainfall
intensity. In order to eliminate ‘noise’ which may interfere in the analysis, the raw
data was initially subjected to autoscaling. Autoscaling involves y-mean scaling
followed by standardisation of the variables. This ensured that all the variables have
equal weights in the analysis (Purcell et al. 2005; Settle et al. 2007; Zhang et al.
2006).
The pre-treated data matrix was then subjected to Hotelling T2 test. This was to
identify atypical samples which are samples that can affect the statistical significance
of the whole data matrix. Hotelling ellipse encompasses all the samples that lie
within 95% confidence interval limit. The samples which were outside the T2
Hotelling ellipse were taken as atypical samples and were removed from the data
matrix prior to further analysis.
Application of Principal Component Analysis (PCA)
As discussed in Section 4.5.2, PCA biplot and the correlation matrix which shows
the degree of correlation among the parameters was used in the analysis in order to
identify the best correlated parameters. When the best correlated parameters are
identified, the potential surrogate indicators for a parameter of interest can be
decided. PCA analysis of this research was carried out with the aid of StatistiXL
Version1.5 software. This software was selected due to its versatility, ease of use and
superior data handling capabilities.
Among the parameters measured in this research, EC, TTU, TSS, TDS, TOC and
DOC can be considered as the easiest to measure set of parameters. Turbidity and
electrical conductivity have on site measurement capability and less rigorous
laboratory procedures (Settle et al. 2007). As described in Chapter 3, several
155
researchers have noted that these parameters have the potential to act as surrogate
parameters for other key water quality parameters such as nitrogen, phosphorus and
heavy metals (Grayson et al. 1996; Han et al. 2006; Herngren 2005; Settle et al.
2007). Therefore, special attention was given to finding correlations for nitrogen and
phosphorus compounds with EC, TTU, TSS, TDS, TS, TOC and DOC in the PCA
analysis. In addition, analysis was also carried out to identify the potential surrogate
parameters for TSS, TDS and TS using EC and TTU.
8.3.1 Identification of potential surrogate parameters for road surfaces
Figure 8.6 shows the PCA analysis biplot for both road surfaces. The PC1 versus the
PC2 biplot accounts for almost 72% of data variance. This indicates that most of the
information is explained by the biplot.
The main purpose of PCA was to identify correlated parameters and to group them
depending on their correlations. Accordingly, the potential surrogate parameters
could be identified for these groups separately.
Figure 8.6 leads to the following conclusions;
• TNO3, DNO3, TNO2, DNO2, TN, DTN, TKN and DKN are strongly correlated to each other (See group 1);
• TPO4, DPO4, TP and DTP are strongly correlated to each other (See group 2) and;
• TSS, TDS and TS are strongly correlated to each other (See group 3).
156
Figure 8.6- PCA biplot for all the physico-chemical parameters for road
surfaces Note: TTU- Turbidity; EC- Electrical conductivity; TNO2- Total nitrite-nitrogen; DNO2- Dissolved
nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3- Total nitrate-
nitrogen; DNO3- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN Dissolved kjeldahl
nitrogen -TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS-
Total dissolved solids; TS- Total solids; TPO4- Total Phosphates; DPO4- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
The correlation matrix which is resultant from PCA shows the degree of correlation
among the parameters (Table 8.2). It is always recommended to confirm the visual
correlation evident in PCA biplots with a correlation matrix (Carroll and
Goonetilleke 2005; Farnham et al. 2003; Rahman et al. 2002). In this study, variables
with a correlation coefficient greater than 0.50 was considered as strongly correlated
parameters. The variables with a correlation coefficient in the range of 0.35-0.50
were considered as parameters with some correlation. As seen in Table 8.2, all
phosphorus compounds (TPO4, DPO4, TP and DTP) are strongly correlated to each
other as they indicate a correlation coefficient of greater than 0.90. Similarly, all
157
nitrogen compounds show strong correlation to each other with correlation
coefficients in the range of 0.35- 0.99.
Table 8.2- Correlation matrix of physico-chemical parameters obtained from principal component analysis
Note: TTU- Turbidity; EC- Electrical conductivity; TNO2- Total nitrite-nitrogen; DNO2- Dissolved
nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3- Total nitrate-
nitrogen; DNO3- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN Dissolved kjeldahl
nitrogen -TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS-
Total dissolved solids; TS- Total solids; TPO4- Total Phosphates; DPO4- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
Based on the observations derived from the PCA biplot and the correlation matrix,
potential surrogate parameters can be identified for the following three groups as
confirmed by Figure 8.6:
Group 1- For all nitrogen compounds;
Group 2- For all phosphorus compounds; and
Group 3- For TSS, TDS and TS.
158
• Identification of potential surrogate parameters for nitrogen compounds
As seen in Figure 8.6, all nitrogen parameters show strong correlation to each other.
TN is the sum of all nitrogen forms. Nitrogen compounds in the wash-off at both
study sites are mostly available in dissolved form. Table 8.4 gives the mean of TKN
and TN measured for both study sites. On average, only 20%- 32% of TKN and TN
are available as particulate nitrogen in both study sites. Previous studies have also
confirmed that dissolved nitrogen is the primary form of nitrogen compounds in
urban stormwater runoff (Taylor et. al. 2005; Uunk and Ven 1987). Uunk and Ven
(1987) in their study found that particulate nitrogen was below 33% of the total
nitrogen in urban runoff, while Taylor et. al. (2005) reported that particulate nitrogen
accounted for only around 20% of the total nitrogen in urban runoff. Therefore, DTN
was selected as the most representative parameter for all nitrogen compounds. The
surrogate parameters identified for DTN would be suitable surrogate parameters for
all other nitrogen compounds.
Table 8.3- Mean concentrations of nitrogen compounds
Note: TKN- Total kjeldahl nitrogen; DKN- Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN-
Dissolved total nitrogen.
For the identification of surrogate parameters for DTN, a separate PCA was carried
out. The PCA biplot which was developed shows the extent of the correlation of
DTN with TSS, TDS, TOC and DOC (Figure 8.7). These parameters were selected
as they are deemed as an easy to measure set of parameters. As TTU and EC were
not correlated with DTN (See Figure 8.6) they were not included in this PCA.
Nitrogen parameter (mg/L) Particulate percentage
(%) Site ID TKN DKN TN DTN TKN TN
Drumbeat Street
2.70 2.13 2.82 2.23 20 20
Ceil Circuit 2.96 2.02 3.13 2.15 32 32
159
TSS
TDS
TS
TOC
DOC
DTN
-4
-3
-2
-1
0
1
2
3
4
-5 0 5 10
PCA 1 (63.3%)
PC
A 2
(25
.5%
)
Figure 8.7- PCA biplot for DTN with easy to measure parameters
Note: TOC- Total organic carbon; DOC- Dissolved organic carbon; DTN- Dissolved total nitrogen;
TSS- Total suspended solids; TDS- Total dissolved solids; TS- Total solids.
As seen in Figure 8.7, DTN is strongly correlated to DOC. According to Table 8.2,
DTN correlation with DOC is also confirmed by the high correlation coefficient of
0.903. Therefore, DOC could be a potential surrogate parameter for DTN.
Furthermore, as seen in Figure 8.7, there is no visible correlation of DTN and TDS
as these two vectors are at nearly 90º. However, interestingly as seen in Table 8.2
these parameters correlated with a correlation coefficient of 0.556. The correlation
matrix represents the evaluation of the whole data set while PCA biplot shown in
Figure 8.7 only shows around 89% of total data variance. Consequently, even though
the degree of correlation between TDS and DTN was not clear in the biplot (Figure
8.7), considering the correlation coefficient of 0.556, TDS was also selected as a
potential surrogate parameter for DTN and the validity of this selection was checked
in the PLS analysis.
160
Also, several researchers have noted that observations obtained from PCA biplots
and correlation matrices should be supported by the raw data matrix (Deb et al.
2008; Kokot et al. 1998; Pommer et al. 2004). Therefore, exploration of the raw data
matrix was also carried out in order to validate the results obtained from PCA. The
variation of DTN with DOC and TDS was plotted as shown in Figure 8.8a, 8.8b,
8.8c, 8.8d. As evident in Figure 8.8a, 8.8b, 8.8c, 8.8d, DTN concentration in each of
the wash-off samples decreases with the decreasing DOC and TDS concentrations.
Hence, TDS and DOC can be considered as potential surrogate indicators for DTN.
This is further supported by the findings of Zeng and Rasmussen (2005) in a lake
monitoring study in USA, who noted that TDS can be used as an indicator
measurement of nitrogen concentrations. Furthermore, past researchers have noted a
strong correlation of DOC with TKN, which is the organic fraction of the nitrogen
compounds (Han et al. 2006; Zeng and Rasmussen 2005).
0
1
2
3
4
5
6
7
8
0 20 40 60 80 100 120 140 160
DOC concentration (mg/L )
DT
N c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.8a- Variation of DTN with DOC for Drumbeat Street
161
0
1
2
3
4
5
6
7
8
0 100 200 300 400 500 600
TDS concentration (mg/L)
DT
N
conc
entr
atio
n(m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.8b- Variation of DTN with TDS for Drumbeat Street
0
1
2
3
4
5
6
7
8
9
0 10 20 30 40 50
DOC concentration (mg/L)
DT
N c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.8c- Variation of DTN with DOC for Ceil Circuit
162
0
1
2
3
4
5
6
7
8
9
0 50 100 150 200 250
TDS concentration (mg/L)
DT
N
conc
entra
tion(
mg/
L)20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.8d- Variation of DTN with TDS for Ceil Circuit
• Identification of potential surrogate parameters for phosphorus
compounds
Phosphorus in stormwater runoff exists in either organic or inorganic form and is
available in particulate or dissolved phases (US EPA 1999). Total phosphorus (TP)
and orthophosphates (PO43-) are the key indicator parameters of phosphorus in urban
stormwater runoff (Lee and Bang 2000; Atasoy et al. 2006). TP is the sum of all
forms of phosphorus compounds. As seen in Figure 8.6, all phosphorus compounds
are strongly correlated to each other. This is further confirmed by Table 8.2, where
all phosphorus compounds show correlation coefficients of more than 0.900.
Among the phosphorus compounds, particulates are dominant at both study sites. On
average, more than 58% are in particulate form (See Table 8.4 below). Past research
studies have noted similar results for runoff from urban impervious surfaces (Atasoy
et al. 2006; Jian-Wei et al. 2007). Jian-Wei et al. (2007) noted that particulate
phosphorus accounted for 66% of total phosphorus in runoff from Wuhan City which
is an urban tourist area in China. Therefore, TP was selected as the representative
parameter of all phosphorus compounds. Incidentally, the surrogate parameters
163
identified for TP would be a suitable parameter for all the other phosphorus
compounds.
Table 8.4- Mean concentrations of phosphorus compounds
Phosphorus parameter (mg/L) Particulate
percentage (%) Site ID TPO43- DPO4
3- TP DTP PO43- TP
Drumbeat Street 0.45 0.19 0.56 0.22 58 60 Ceil Circuit 0.20 0.08 0.28 0.10 62 64
Note: TPO43-- Total Phosphates; DPO4
3-- Dissolved Total Phosphates; TP- Total phosphorus; DTP-
Dissolved total phosphorus.
In order to understand the correlation of TP with an easy to measure set of
parameters, PCA was carried out. As evident in Figure 8.6, TTU was not correlated
to TP. Furthermore, it was noted that particulate phosphorus is the dominant form of
phosphorus. Consequently, EC, TDS, TTU and DOC were not included in this PCA.
PCA was carried out for TP with the remaining set of easy to measure parameters,
namely, TSS, TS and TOC. The resulting PCA biplot is shown in Figure 8.9. The
PCA biplot obtained accounts for almost 90% of the data variance which indicates
that most of the information is explained by the biplot.
164
TSS
TS
TOC
TP
-4
-3
-2
-1
0
1
2
3
-5 0 5 10
PCA 1 (69.6%)
PC
A 2
(18
.9%
)
Figure 8.9- PCA biplot for TP with easy to measure parameters Note: TOC- Total organic carbon; TSS- Total suspended solids; TS- Total solids; TP- Total
phosphorus.
According to Figure 8.9, TOC and TS can be identified as being the most closely
correlated parameters for TP. Furthermore, TOC and TS has correlation coefficients
of 0.757 and 0.693 with TP respectively, confirming the high correlation of these
parameters (See Table 8.2). Furthermore, it was noted that for both study sites, TP
decreases with the decrease in TOC and TS concentrations (See Figure 8.10a, 8.10b,
8.10c, 8.10d).
Novotny (1995) found that phosphorus is distributed in stormwater runoff as a solids
bound pollutant. They suggested that the behaviour of phosphorus is highly
influenced by the solids concentration. Mallin et al. (2008) found a high correlation
of phosphorus with TSS after investigating stormwater runoff from urban impervious
surfaces in two countries in South East of USA. On the other hand, several
researchers have found that, particulate organic matter was highly correlated to
165
particulate phosphorus (Krongvang 1992; Walling and Kane 1982) Consequently,
TOC and TS can be considered as potential surrogate indicators for TP in urban
stormwater runoff.
0.0
0.5
1.0
1.5
2.0
2.5
0 50 100 150 200TOC concentration (mg/L)
TP
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.10a- Variation of TP with TOC for Drumbeat Street
0.0
0.5
1.0
1.5
2.0
2.5
0 100 200 300 400 500 600 700
TS concentration (mg/L)
TP
con
cetr
atio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.10b- Variation of TP with TS for Drumbeat Street
166
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0 10 20 30 40 50
TOC concentration (mg/L)
TP
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.10c- Variation of TP with TOC for Ceil Circuit
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0 50 100 150 200 250 300 350
TS concentration (mg/L)
TP
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.10d- Variation of TP with TS for Ceil Circuit
167
• Identification of potential surrogate parameters for TSS, TDS and TS
As discussed in Chapter 2, solids occur in stormwater runoff as in dissolved or
particulate forms. Solids are the most important stormwater runoff pollutant in terms
of load and have additional significance, because other pollutants such as
hydrocarbons, heavy metals and nutrients are attached to the solids (Deletic et al.
1998; Herngren 2005). Consequently, investigation of solids in runoff is important in
providing stormwater pollution mitigating actions. In this context, measurement of
solids in urban stormwater runoff is significant.
As discussed in Chapter 6, TSS, TDS and TS are the key indicator parameters of
solids present in urban stormwater runoff. These are relatively easy to measure
parameters in comparison with parameters such as TKN and TP. However, there are
monitoring programs which are focused on measuring solid concentrations and the
transport rate of solids, which can vary rapidly (Lewis 1996). In this context, the
frequency of data collection is important (Deletic et al. 1998). In such situations,
determining solids concentrations is often impractical and expensive. Instead, the
identification of easy to measure surrogate parameter/s for solids is important.
Therefore, the investigation of parameters such as EC and TTU as surrogate
parameters for TSS, TDS and TS should be considered.
Figure 8.11, shows clearly the correlation of TSS, TDS and TS with EC and TTU.
The degree of correlation of these parameters was observed from the correlation
matrix (Table 8.2).
168
TS
TDS
TSS
TTU
EC
-2
-1
0
1
2
3
4
5
-4 -2 0 2 4 6
PCA 1 (58.4%)
PC
A 2
(20
.5%
)
Figure 8.11- Correlation of TSS, TDS and TS with EC and TTU Note: TTU- Turbidity; EC- Electrical conductivity; TSS- Total suspended solids; TDS- Total dissolved solids; TS- Total solids.
According to Figure 8.11, TSS shows a correlation with TTU with a correlation
coefficient of 0.553 (See Table 8.2). Figure 8.12a, 8.12b show the variation of TSS
with TTU for both study sites. As evident in these figures, TSS decreases with
decreasing TTU for all the intensities. This further confirms the extent of correlation
between these parameters. Therefore, TTU could be a potential surrogate parameter
for TSS in urban stormwater runoff. According to Lewis (1996), turbidity is a good
measure of solids in stormwater runoff. Furthermore, turbidity is a measure of the
attenuation or scattering of a light beam by particulate and dissolved solids in a water
column (Packman et al. 1999). Therefore, turbidity has the potential to provide the
most direct measure of particulate concentration of solids. This offers an approach
which is applicable to direct field based measurement of TSS. This approach reduces
the time associated with measurement of TSS (APHA 2005). Also, measuring
turbidity is one of the least expensive and easiest methods. Additionally, past
researchers have noted the use of turbidity as a potential surrogate indicator for
169
solids in urban stormwater runoff (Deletic et al. 1998; Grayson et al. 1996; Settle et
al. 2007).
0
50
100
150
200
250
0 20 40 60 80 100TTU concentration (NTU)
TS
S c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.12a- Variation of TSS with TTU for Drumbeat Street
0
20
40
60
80
100
120
0 5 10 15 20 25 30 35 40
TTU concentration (NTU)
TS
S c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.12b- Variation of TSS with TTU for Ceil Circuit
170
However, the correlation between these indicators will always be approximate due to
a number of reasons. A higher TSS for a given turbidity may be explained by higher
concentrations of the fine fraction of solids or higher concentrations of fine
particulate organic matter. According to Gippel (1995), mineral particles contained
in solids can cause lower turbidity levels in water samples. This suggests that, the
correlation between these indicators will always vary with the influence of particle
shape, size and amount of surface area of solid particles which can cause variations
in reflection, refraction and absorption of light (Packman et al. 1999). According to
Packman et al. (1999), TSS and turbidity relationship can be affected by water
colour, because of the dissolved organic compounds which can absorb more light
than inorganic compounds.
As evident in Figure 8.11 and Table 8.2, TDS shows some correlation with EC with
a correlation coefficient of 0.436. The correlation between TDS and EC suggests the
possibility of considering EC as a surrogate for TDS. TDS represents the total
quantity of dissolved solids, which are both organic and inorganic forms of
pollutants. These represent both positively and negatively charged ions. On the other
hand, EC is a measure of the number of charged particles. Consequently, EC can be
considered as a potential indicator of total dissolved solids in stormwater (Chapman
1992; Settle et al. 2007; Zeng and Rasmussen 2005). Use of EC as an indicator for
TDS means that direct measurements in the field can reduce the time and cost
associated with the laboratory measurement of TDS.
This relationship is well documented in several research studies (Settle et al. 2007;
Zeng and Rasmussen 2005). According to Chapman (1992), TDS concentration may
be obtained by multiplying EC value by a factor which is commonly between 0.55
and 0.75. However, the correlation of dissolved particles to conductivity is
influenced by organic matter content and other pollutants such as hydrocarbons.
Atekwana et al. (2004) in their study found that the reduction in electrical
conductivity was due to the presence of hydrocarbons which may have the potential
to affect the dissolved solids predicted to be in ground water.
171
Total solids (TS) are a measure of both suspended solids and total dissolved solids in
stormwater runoff. There are monitoring programs where direct measurement of TS
is needed rather than taking separate measurements of TSS and TDS. In such cases,
identification of separate parameters for TS as surrogates is important. As seen in
Figure 8.11 and Table 8.2, TS shows some correlation with TTU and EC with a
correlation coefficient of 0.370 and 0.414 respectively. As discussed above, TTU
and EC are potential surrogate parameters for TSS and TDS respectively. Hence,
there is a possibility of using the parameters TTU and EC as surrogates for TS.
8.3.2 Identification of potential surrogate parameters for roof surfaces
Similar to the road surfaces, analysis of physico-chemical parameters was carried out
for the roof surface runoff. Figure 8.13 shows the PCA analysis biplot. The PC1
versus PC2 biplot accounts for almost 70% of data variance. The main purpose of
the PCA was to identify the correlated parameters and group them depending on
their correlations. Consequently, the surrogate parameters were identified for those
groups separately. The correlation matrix which is resultant from PCA shows the
degree of correlation among the parameters (Table 8.5). Similar to the road surfaces,
parameters which show a correlation coefficient greater than 0.50 was considered as
strongly correlated parameters and parameters with 0.35-0.50 was considered as
having some correlation.
172
Figure 8.13- PCA biplot for all the physico-chemical parameters for roof surfaces
Note: TTU- Turbidity; EC- Electrical conductivity; TNO2- Total nitrite-nitrogen; DNO2- Dissolved
nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3- Total nitrate-
nitrogen; DNO3- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN- Dissolved kjeldahl
nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS-
Total dissolved solids; TS- Total solids; TPO4- Total Phosphates; DPO4- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
Figure 8.13 leads to the following conclusions:
• TNO3, DNO3, TNO2, DNO2, TN, DTN, TKN and DKN are strongly correlated to each other (See group 1);
• TPO4 ,DPO4, TP and DTP are strongly correlated to each other (See group 2); and
• TSS, TDS and TS are strongly correlated to each other (See group 3).
173
Table 8.5- Correlation matrix obtained from PCA
Note: TTU- Turbidity; EC- Electrical conductivity; TNO2- Total nitrite-nitrogen; DNO2- Dissolved
nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3- Total nitrate-
nitrogen; DNO3- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN- Dissolved kjeldahl
nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS-
Total dissolved solids; TS- Total solids; TPO4- Total Phosphates; DPO4- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
Based on the conclusions derived from the PCA biplot and the correlation matrix,
potential surrogate parameters can be identified for the following three groups:
• For nitrogen compounds;
• For phosphorus compounds; and
• For TSS, TDS and TS.
• Identification of potential surrogate parameters for nitrogen compounds
All nitrogen compounds show good correlation to each other. According to several
research findings, TN in roof surface runoff is a significant stormwater pollutant
(Gobel et al. 2006; Huang et al. 2007; Thomas and Greene 1993). Exploring the raw
data matrix as seen in Table 8.6, the dissolved fraction of TN is the dominant form of
nitrogen in roof surface runoff which is around 80%. Consequently, DTN was
selected as the most representative parameter for all nitrogen compounds. The
174
surrogate parameters identified for DTN would be the surrogate parameters for all
nitrogen compounds.
Table 8.6- Mean concentrations of nitrogen compounds
Note: TKN- Total kjeldahl nitrogen; DKN Dissolved kjeldahl nitrogen -TN- Total nitrogen; DTN-
Dissolved total nitrogen.
PCA was carried out to assess the correlation between DTN and the set of easy to
measure parameters. Figure 8.14 shows the biplot obtained for DTN with EC, TTU,
TOC, DOC, TSS, TDS and TS. As seen in Figure 8.14, DTN was strongly correlated
to TDS. Therefore, TDS was selected as the best indicator of dissolved total
nitrogen. This selection was further supported by the correlation coefficient of TDS
and TN which is 0.503 (See Table 8.5).
EC
TTU
TSS
TDS
TS
TOC
DOC
DTN
-5
-4
-3
-2
-1
0
1
2
3
-4 -2 0 2 4 6
PCA 1 (44.2%)
PC
A 2
(31
.4%
)
Figure 8.14- PCA biplot for DTN with easy to measure parameters Note: TTU- Turbidity; EC- Electrical conductivity; TOC- Total organic carbon; DOC- Dissolved
organic carbon; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS- Total dissolved
solids; TS- Total solids.
Nitrogen parameter (mg/L) Percentage in the
dissolved fraction (%) Site ID TKN DKN TN DTN TKN TN Roofs 0.54 0.44 0.70 0.57 80 81
175
Figure 8.15 shows the variation of DTN concentration with TDS concentration for
the raw data matrix. As seen in Figure 8.15, DTN decreases with decreasing TDS
concentrations for all the intensities. Hence, TDS can be considered as a potential
surrogate parameter for DTN in roof surface runoff.
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
0 20 40 60 80 100 120
TDS concentration (mg/L)
DT
N c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.15- Variation of DTN with TDS for roof surfaces
• Identification of potential surrogate parameters for phosphorus
compounds
As discussed in Chapter 2, phosphorus compounds are significant stormwater
pollutants. However, though research literature is available on the investigation of
phosphorus compounds on road surface runoff, investigation of phosphorus
compounds in roof surface runoff is scare (Chang and Crowley 1993; Polkowska et
al. 2002).
According to Figure 8.13 and Table 8.5, all phosphorus compounds are strongly
correlated to each other with correlation coefficients of greater than 0.750. TP is the
sum of all forms of phosphorus. Exploring the raw data matrix, it was noted that
around 65% of total phosphorus is in particulate form (See Table 8.7). Therefore, TP
176
can be considered as the indicator parameter for phosphorus in urban roof surface
runoff. However, the surrogate parameters identified for TP would be common
surrogate parameters for other phosphorus compounds.
Table 8.7- Mean concentrations of phosphorus compounds
Phosphorus parameter (mg/L) Particulate
percentage (%) Site ID TPO43- DPO4
3- TP DTP PO43- TP
Roof surface 1.80 0.64 1.85 0.64 65% 66% Note: TPO43-- Total Phosphates; DPO43-- Dissolved Total Phosphates; TP- Total phosphorus; DTP-
Dissolved total phosphorus.
In order to have an understanding of the correlation of TP with the easy to measure
set of parameters, PCA was carried out for TP with TTU, TOC, TSS, and TS. As
phosphorus is in particulate form, EC, DOC and TDS were not included in this PCA.
The resulting biplot is shown in Figure 8.16. According to Figure 8.16, TTU and
TOC show negative correlation with TP. According to Table 8.5, correlation
coefficient of TP with TTU and TOC are-0.541 and -0.403 respectively.
TP
TOC
TS
TSS
TTU
-4
-3
-2
-1
0
1
2
3
-4 -2 0 2 4 6
PCA 1 (49.1%)
PC
A 2
(35
.2%
)
Figure 8.16- PCA biplot for TP with easy to measure parameters Note: TTU- Turbidity; TOC- Total organic carbon; TSS- Total suspended solids; TS- Total solids;
TP- Total phosphorus.
177
However, exploring the raw data matrix, this negative correlation was not evident for
all the intensities. As seen in Figure 8.17a, 8.17b, TP concentration decreases with
decreasing TOC and TTU concentrations which suggests positive correlation among
the parameters. This pattern of variation was contradictory to the observations noted
in PCA analysis. Therefore, identification of surrogate parameters for phosphorus
was not successful.
0
1
2
3
4
5
6
0 2 4 6 8 10 12TOC concentration (mg/L)
TP
con
cent
ratio
n(m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.17a- Variation of TP with TOC
178
0
1
2
3
4
5
6
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
TTU (NTU)
TP
con
cent
ratio
n(m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.17b- Variation of TP with TTU
• Identification of potential surrogate parameters for TSS, TDS and TS
Similar to the road surface runoff, concentration of solids in roof surface runoff is
important as the particulates can carry a variety of pollutants such as nutrients and
heavy metals into receiving water bodies (Forster 1999; Gadd and Kennedy 2001;
Gnecco et al. 2005). In this context, the identification of surrogate parameters for
solids is important as it can provide a convenient method to measure TSS, TDS and
TS which are the key indicators of solids in roof runoff, based on simple field
measurements such as EC and TTU. Therefore, PCA was carried out for TSS, TDS,
TS with EC and TTU.
Figure 8.18 shows the PCA biplot obtained for TSS, TDS and TS with EC and TTU.
The PCA biplot accounts for almost 90% of data variance which indicates that most
of the information is explained by the biplot. Correlation matrix given in Table 8.5
shows the degree of correlation of these parameters. It was noted that TSS is strongly
correlated to EC and TDS is strongly correlated to TTU with correlation coefficients
of 0.585 and 0.504 respectively. However, the correlation of TSS with EC and TDS
with TTU are contradictory to the findings of several researchers who noted that EC
179
and TTU as potential surrogate indicators for TDS and TSS respectively (Gippel
1995; Zeng and Rasmussen 2005).
Exploration of the raw data matrix was carried out in order to have a clear
understanding of these correlations (See Figure 8.19a, 8.19b, 8.19c, 8.19d). The
correlation of TSS with EC and TDS with TTU were not clear in the raw data
matrix. Therefore, identification of potential surrogate parameters for TSS and TDS
was not successful for roof surface wash-off.
TS
TDS
TSS
TTU
EC
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
-5 0 5 10
PCA 1 (64.2%)
PC
A 2
(24
.7%
)
Note: TTU- Turbidity; EC- Electrical conductivity; TSS- Total suspended solids; TDS- Total
dissolved solids; TS- Total solids.
Figure 8.18- PCA biplot for TS, TTU and EC
However, as seen in Figure 8.18 and Table 8.5, TS shows limited correlation to EC
and TTU with correlation coefficients of 0.413 and 0.463 respectively. As evident in
Figure 8.19c and Figure 8.19d, TS decreases with decreasing EC and TTU for most
180
of the intensities. Therefore, both EC and TTU were considered as potential
surrogate indicators for TS in roof surface runoff.
0
10
20
30
40
50
60
70
80
90
100
35 45 55 65 75 85 95
TS
S c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
EC (µS/cm)
Figure 8.19a- Variation of TSS with EC
0
20
40
60
80
100
120
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
TTU (NTU)
TD
S c
once
ntra
tion
(mg/
L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.19b- Variation of TDS with TTU
181
0
50
100
150
200
250
25 35 45 55 65 75 85 95
TS
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
EC (µS/cm)
Figure 8.19c- Variation of TS with EC
0
50
100
150
200
250
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
TTU (NTU)
TS
con
cent
ratio
n (m
g/L)
20 mm/hr
40 mm/hr
65 mm/hr
86 mm/hr
115 mm/hr
135 mm/hr
Figure 8.19d- Variation of TS with TTU
182
Table 8.8 presents the summary of the potential surrogate parameters which were
obtained for nitrogen compounds, phosphorus compounds and solids in the runoff
from both road and roof surfaces.
Table 8.8- Potential surrogate water quality parameters for nitrogen, phosphorus and solids
Constituent Key indicator Potential surrogate
parameter
Roads
Nitrogen Dissolved total nitrogen(DTN)
Total dissolved solids (TDS)
Dissolved organic carbon (DOC)
Phosphorus Total phosphorus (TP) Total solids (TS)
Total organic carbon (TOC)
Solids Total suspended solids
(TSS)
Total dissolved solids (TDS)
Total solids (TS)
Turbidity(TTU)
Electrical conductivity(EC)
Turbidity (TTU)
Electrical conductivity(EC)
Roofs
Nitrogen Dissolved total nitrogen(DTN)
Total dissolved solids (TDS)
Phosphorus Total phosphorus (TP) Not found
Solids Total suspended solids(TSS)
Total dissolved solids (TDS)
Total solids (TS)
Not found
Not found
Electrical conductivity(EC)
Turbidity (TTU)
183
8.4 Verification of selected surrogate parameters using PLS
Section 8.3 discussed the potential surrogate parameters for nitrogen, phosphorus
and solids in wash-off from both road and roof surfaces. It was needed to check the
suitability of the selected parameters as surrogate parameters. Therefore, to assess
the validity of this selection, PLS regression was used as discussed in Section 8.1.
PLS analysis was carried out with the aid of PRS Serius 7.1 software.
Detailed description of PLS model development and validation is available in
Section 4.5.2. As discussed in Section 4.5.2, mathematically, PLS relates one or
more response variables (denoted Y) to two or more predictor variables (denoted X)
in the model development. Therefore, PLS model development was not performed
where only one parameter is acting as a surrogate parameter. For example, as seen in
Table 8.8, only TDS was found as a surrogate parameter for DTN in roof surface
runoff and no model was developed for that. Furthermore, for each parameter set a
number of different models were tested by using different calibration and validation
data sets.
As discussed in Section 4.5.2, SECV, R2, SEP and r2 are common statistics which
are used in PLS (Ayoko et al. 2007; Goonetilleke et al. 2004; Herngren 2005). Low
SECV with high R2 indicates the excellent validity of the calibration model. The best
calibration is the one with the highest coefficient of determination (r2) and the lowest
Standard Error of Performance (SEP). However, researchers have noted that the use
of only of these statistics can be misleading (Batten 1998; Campbell et al. 1997;
Dunn et al. 2002; Williams 1987). Therefore, they have suggested the use of the ratio
of the standard deviation of the Y variable in the validation set to SEP which is
known as RPD [Ratio of (standard error of) Performance to (standard) Deviation]
which can give additional information which describes the quality of the model. For
example, considering model 1in Table 8.9, RPD was calculated by dividing the
standard deviation of DTN in validation set (32.446), by the SEP values (9.80). This
resulted in a RPD of 3.3 for model 1.
184
According to several research findings, no critical level for RPD is defined. The
acceptable value depends on the intended application of the predictive values.
According to Malley et al. (1999) in agricultural applications RPD>3 is considered
as acceptable and RPD>5 is considered as excellent. Chang et al. (2001) reported
that in the case of soil properties, RPD in the range of >2, 1.4-2.0 and <1.4 indicate
decreasing reliability of prediction. Furthermore, Dunn et al. (2002) noted that
suitable limits for RPD as <1.6 as poor, 1.6-2.0 as acceptable; and. >2.0 as excellent
in the analysis of soils for site specific agriculture. However, use of RPD value to
describe the PLS models in research relating to water quality is rare. Therefore, in
this research, models with RPD>3 were taken as good and RPD value in the range of
1.4-3.0 were taken as acceptable. Table 8.9 summaries the results of the models
which were selected as the best model from the number of models developed for
each parameter set using different calibration and validation data sets.
185
Table 8.9- Calibration and prediction results of models
Model Response variable (Y)
Predictor variables (X)
No of samples in cal.
Calb. R2
SECV No of samples in validation
Pred. r2 SEP RPD
Roads
1. DTN DTN TNO2, DNO2, TNO3, DNO3, TKN, DTKN, TN
32 0.96 5.29
32 0.93 9.80 3.3
2. DTN-surrogate
DTN TDS, DOC 32 0.77 14.74 32 0.94 8.28 4.0
3. TP TP TPO4, DPO4, DTP 32 0.92 6.95
32 0.98 3.52 7.0
4. TP-surrogate
TP TS, TOC 32 0.61 14.40
32 0.60 16.07 1.6
5. TS-surrogate
TS EC, TTU 32 0.30 12.41
32 0.65 13.79 1.8
Roofs
6. DTN DTN TNO2, DNO2, TNO3, DNO3, TKN, DTKN, TN
16 0.98 4.255
16 0.97 5.433 5.2
7. TP TP TPO4, DPO4, DTP 16 0.99 26.456 16 0.99 27.41 10.4
8. TS-
surrogate
TS EC, TTU 32 0.60 39.23
16 0.79 - -
186
Following is a brief description of the models which are shown in Table 8.9.
• 1 and 6- DTN models - These models were developed to check the validity of
the selection of DTN as the most representative parameter of all nitrogen
compounds for road surfaces and roof surfaces. In these models, DTN was
taken as response (Y) variable and all other nitrogen parameters namely,
TNO2, DNO2, TNO3, DNO3, TKN, DKN and TN were taken as predictor- X
variables.
• 2- DTN surrogate model - This model was developed to check the validity of
the selected surrogate parameters namely, TDS and DOC for DTN for road
surfaces. In this model, DTN was taken as Y variable. TDS and DOC were
taken as X variables.
• 3 and 7- TP model - These models were developed to check the validity of the
selection of TP as the most representative parameter of all phosphorus
compounds for road surfaces and roof surfaces. In these models, TP was taken
as Y variables and all other phosphorus parameters namely, TPO4, DPO4 and
DTP were taken as X variables.
• 4- TP surrogate model - This model was developed to check the validity of the
selection of surrogate parameters namely, TS and TOC for TP for road
surfaces. In this model, TP was taken as Y variable, while, TS and TOC were
taken as X variables; and
• 5 and 8- TS surrogate model - These models were developed to check the
validity of selected surrogate parameters, namely, EC and TTU for TS for road
surfaces and roof surfaces. In these models, TS was taken as Y variable and EC
and TTU were taken as X variables.
The following conclusions were derived from Table 8.9. For roads • Model 1 has good ability to represent all nitrogen compounds in urban
stormwater runoff. The average prediction error is close to 8% with r2= 0.93
and SEP=9.80 and RPD of 3.3.
187
• Model 3 has good ability to represent all phosphorus compounds in urban
stormwater runoff. The model has a r2=0.98 which is the highest coefficient of
determination and SEP of 3.52 which is the lowest SEP in comparison to the
other models relating to roads. RPD value which is 7 further confirms the
applicability of this model.
• Model 2 has good ability to predict DTN from TDS and TOC with r2 of 0.94
and SEP of 8.28 and RPD of 4.0.
• According to model 4, predictive ability of TP from TS and TOC is reasonable
with r2=0.60 and SEP of 16.07 and RPD of 1.6.
• According to model 5, prediction of TS by EC and TTU is reasonable with r2=
0.65, SEP of 13.79 and RPD of 1.8.
For roofs
• Model 6 and 7 show good ability to represent all nitrogen and phosphorus
compounds in roof surface runoff with DTN and TP respectively. These
models show high RPD values of 5.2 and 10.4 respectively.
• Model 8, which predicts TS from EC and TTU perform with r2=0.78 is a
relatively weak model due to the high SECV value. As model 8 was developed
from the full data set, SEP and RPD values are not available for evaluating the
quality of the model. However, the confidence of the prediction TS from EC
and TTU could be further understand by analysing a relatively larger data set.
8.5 Conclusions
The following important conclusions were derived from the analysis of pollutant
wash-off data from road and roof surfaces.
• Pollutant wash-off investigated in this study was in agreement with the general
understanding of the wash-off process in research literature.
188
• DTN and TP can be considered as the primarily available nitrogen and
phosphorus forms in wash-off from both road surfaces and roof surfaces.
Therefore, these parameters can be used to represent all nitrogen and
phosphorus compounds in stormwater runoff.
• The study has identified the following surrogate parameters to be used in
determining nitrogen, phosphorus and solids in road surface runoff:
• For nitrogen compounds- TDS and DOC
• For phosphorus compounds- TS and TOC
• For TSS- Turbidity
• For TDS- EC
• For TS- Turbidity and EC
• The study has derived the following surrogate parameters to be used in
determining nitrogen, and solids in roof surface runoff:
• For nitrogen compounds- TDS
• For TS- EC and Turbidity
Surrogate parameters with a reasonable level of accuracy for phosphorus
compounds, total suspended solids and total dissolved solids in roof surface runoff
were not found.
189
Chapter 9 - Development of Surrogate Parameter Relationships and Validation
9.1 Background
Chapter 8 discussed the potential surrogate parameters for nitrogen, phosphorus and
solids in urban stormwater runoff from road and roof surfaces. Clear mathematical
relationships between key parameters and their surrogate parameters can
significantly enhance the efficiency of stormwater quality monitoring programs
(Robien et al. 1997; Settle et al. 2007; Thomson et al. 1997). This is mainly
attributed to the fact that these relationships reduce the number of key water quality
parameters to be monitored in a monitoring program. In stormwater quality
monitoring programs, a shorter list of water quality parameters can potentially
reduce analytical costs substantially (Kayhanian et al. 2007).
This Chapter describes the development of statistically based relationships between
the key parameters of interest and the selected surrogate parameters which were
discussed in Chapter 8. Development of relationships was carried out using linear
regression analysis (Robien et al. 1997; Thomson et al. 1997). The portability of the
developed relationships is investigated in this Chapter in order to provide an
understanding of the applicability of relationships at sites other than those for which
the relationships were derived. Finally, this chapter provides a set of surrogate
parameter relationships which can be applied directly to evaluate stormwater quality.
9.2 Development of parameter relationships
The mathematical relationships between the key parameters of interest and the
selected surrogate water quality parameters were developed using regression analysis
including linear, log and power relationships. Several researchers have noted that the
log relationships do not explain as much of the variance as the linear relationships
(Robien et al. 1997; Thomson et al. 1997). On the other hand, considering easy
190
applicability, linear regression relationships are widely used in water quality research
studies in comparison to log and power relationships (Robien et al. 1997; Settle et al.
2007; Thomson et al. 1997). Therefore, in this research only linear relationships
were derived between the key parameters of interest and the selected surrogate
parameters. The linear regression analysis was performed using StatistiXL version
1.5 software (Roberts and Withers 2004).
Linear regression explains the variation of one variable which is the response
variable (Y), in terms of one or more predictor variables (X1, X2, ………..Xm). It
assumes that a linear relationship exists between the response variable and the
predictor variable(s) of the form Y= a + bX, where a is the intercept and b is the
slope. As discussed in Chapter 8, for the road and roof surfaces, all the selected
surrogate parameters and their key parameters were positively correlated to each
other. This indicates that when one parameter increases, the other parameter also
increases and visa versa. Furthermore, if a parameter was selected as a surrogate for
a key parameter of interest, it suggests that when surrogate parameter has zero value
its key indicator should also be of zero value. Therefore, the relationships derived
were in the form of Y= bX (Roberts and Withers 2004). For this purpose, regression
was forced through the origin by selecting the Constant=0 option in the regression
analysis (Roberts and Withers 2004).
In this study, relationships developed were in the form of simple linear regression.
Simple linear regression describes the relationship between a single predictor
variable (X) and a single response variable (Y). The goodness of fit of the
relationships derived can be explained using statistical indicators such as coefficient
of determination (R2) and standard error of estimate (SEE) to the mean (Packman et
al. 1999; Robien et al. 1997; Roberts and Withers 2004; Settle et al. 2007; Thomson
et al. 1997).
The coefficient of determination (R2) is the fraction of variability in the response
variable Y that is explained by the variability in the predictor variable(s) X. R2
ranges from 0 (where no variation is explained) to 1 (where all variation is
explained). The standard error of estimate (SEE) is an overall indication of how well
the regression relationship predicts the response of Y to the predictor variables. The
191
smaller the value, the higher the accuracy of the relationship. Therefore, a good
predictive relationship is indicated by high R2 with low SEE. The scatter plot gives a
good estimation of the relationship derived. Scatter plot is a graph which represents
the predictor X variables on the X axis and the response Y variables on the Y axis.
The regression line in a scatter plot is often included with ± 1 standard error or ±95%
confidence limit (Roberts and Withers 2004).
9.2.1 Surrogate parameter relationships for wash-off from road surfaces
Chapter 8 identified surrogate parameters for key water quality parameters for wash-
off from road surfaces. Table 9.1 and Figure 9.1a, 9.1b, 9.1c, 9.1d, 9.1e, 9.1f, 9.1g,
9.1h show the linear regression relationships developed for these parameters. The
units of each parameter are shown with each equation separately. Coefficient of
determination (R2) for each relationship is also shown in Table 9.1. The standard
error of estimate (SEE) is shown in Table 9.1 as a percentage of the mean of each
response variable.
192
Table 9.1- Surrogate parameter relationships for road surfaces
Key parameter
(Y)
Relationship number
Surrogate parameter (X)
Relationship Y = mX
Coefficient of determination
R2
Number of data points (N)
Standard error of
estimate to the mean Y SEE (%)
1 TDS DTN (mg/L) = 0.013TDS (mg/L) 0.82 52 49 DTN
2 DOC DTN (mg/L) = 0.138DOC (mg/L) 0.92 63 34
3 TS TP (mg/L) = 0.002TS (mg/L) 0.84 57 48 TP
4 TOC TP (mg/L) = 0.020TOC (mg/L) 0.86 59 45
TSS 5 TTU TSS (mg/L) = 1.982TTU (NTU) 0.82 60 58
TDS 6 EC TDS (mg/L) = 2.195EC (µS/cm) 0.82 57 48
7 EC TS (mg/L) = 2.735EC (µS/cm) 0.86 58 43 TS
8 TTU TS (mg/L) = 14.281TTU (NTU) 0.90 52 36
193
y = 0.0131xR2 = 0.82
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
0 100 200 300 400
TDS (mg/L)
DT
N (
mg/
L)
Figure 9.1a- Relationship of DTN and TDS
y = 0.138xR2 = 0.92
0
1
2
3
4
5
6
7
8
0 20 40 60
DOC (mg/L)
DT
N (
mg/
L)
Figure 9.1b- Relationship of DTN and DOC
194
y = 0.002xR2 = 0.84
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 200 400 600TS (mg/L)
TP
(m
g/L)
Figure 9.1c- Relationship of TP and TS
y = 0.020xR2 = 0.86
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 20 40 60TOC (mg/L)
TP
(m
g/L)
Figure 9.1d- Relationship of TP and TOC
195
y = 1.982xR2 = 0.82
0
20
40
60
80
100
120
0 20 40 60 TTU( mg/L)
TS
S (m
g/L)
Figure 9.1e- Relationship of TSS and TTU
y = 2.195xR2 = 0.82
0
50
100
150
200
250
300
0 50 100 150EC (µS/cm)
TD
S (
mg/
L)
Figure 9.1f- Relationship of TDS and EC
196
y = 2.735xR2 = 0.86
0
50
100
150
200
250
300
350
400
450
0 50 100 150EC (µS/cm)
TS
mg/
L
Figure 9.1g- Relationship of TS and EC
y = 14.281xR2 = 0.90
0
50
100
150
200
250
300
350
0 10 20 30 TTU (NTU)
TS
(m
g/L)
Figure 9.1h- Relationship of TS and TTU As evident in Table 9.1 and Figure 9.1a, 9.1b, 9.1c, 9.1d, 9.1e, 9.1f, 9.1g, 9.1h, the
DTN-DOC relationship explains good predictability with R2 of 0.92 and SEE of 34%
for the mean of DTN. The TS-TTU relationship shows good predictability of TS
with R2 of 0.90 and SEE of 36% for the mean of TS. Relationships, DTN-TDS, TP-
TS, TP-TOC, TSS-TTU, TDS-EC and TS-EC have reasonable prediction accuracy
197
with SEE of 45%-58% for the mean of each response variable. Comparing the
relationships of TP-TS and TP-TOC, TP-TOC relationship is recommended for use
due to the relatively lower SEE value compared to the TP-TS relationship.
9.2.2 Surrogate parameter relationships for wash-off from roof surfaces
Similar to the road surfaces, predictive relationships were derived for roof surface
runoff for the surrogate parameters identified in Chapter 8. The results of the linear
regression relationships developed are shown in Table 9.2 and Figure 9.2. The units
of each parameter are shown with each equation separately and the standard error of
estimate is indicated as a percentage of the mean of each response variable. R2 is also
given in Table 9.2.
198
Table 9.2- Surrogate parameter relationships for roof surfaces
Key parameter
(Y)
Relationship number
Surrogate parameter
(X)
Relationship Y = mX
Coefficient of determination
R2
Number of data points
(N)
Standard error of estimate to the mean Y
(%)
DTN 1 TDS DTN (mg/L) = 0.011TDS (mg/L) 0.92 31 31
TP No surrogates were found
2 EC TS (mg/L) = 0.759EC (µS/cm) 0.83 26 45 TS
3 TTU TS (mg/L) = 10.640TTU (NTU) 0.74 28 56
TSS No surrogates were found
TDS No surrogates were found
199
y = 0.0112xR2 = 0.92
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 50 100 150TDS ( mg/L)
DT
N (
mg/
L)
Figure 9.2a- Relationship of DTN and TDS
y = 0.759xR2 = 0.83
0
10
20
30
40
50
60
70
80
90
0 20 40 60 80
EC (µS/cm )
TS
( m
g/L)
Figure 9.2b- Relationship of TS and EC
200
y = 10.64xR2 = 0.74
0
20
40
60
80
100
120
0 5 10TTU (NTU)
TS
(m
g/L)
Figure 9.2c- Relationship of TS and TTU As evident in Table 9.2 and Figure 9.2a, 9.2b, 9.2c, the DTN-TDS relationship
shows good predictability of DTN with R2 of 0.92 and SEE of 31% to the mean of
DTN. The prediction of TS from EC and TTU is reasonable with SEEs of 45% and
56% respectively for the mean of TS. Comparing the relationships of TS-EC and TS-
TTU, TS-EC relationship is recommended for use due to the relatively lower SEE
value compared to the TS-TTU relationship.
9.3 Portability of the relationships
Portability refers to the degree to which the developed surrogate parameter
relationships are applicable for the prediction of key water quality parameters using a
separate data set other than the data set which was used to derive the relationships. In
relation to portability, two aspects referred to as ‘near site portability’ and ‘far site
portability’ needs to be considered.
Near site portability refers to the portability of the developed surrogate parameter
relationships to sites which are located in close geographical proximity to the site
which was used to develop the relationships. In this context, environmental
201
differences such as climatic changes and road maintenance practices between the
investigated sites and selected sites are considered to be minimal. Far site portability
refers to the application of the developed relationships to sites where appreciable
meteorological and geographical differences prevail (Hallberg 2006; Kayhanian et
al. 2007; Robien et al. 1997; Thomson et al. 1997).
In this study, the relationships developed were tested only for near site portability
and only for road surfaces, due to the availability of appropriate data sets. For this
purpose, a data set which contains data from three road surfaces namely, Armstrong
Drive, Stevens Street and Lawrence Drive which were also in the Gold Coast was
used. These study sites represent typical characteristics of residential, light industrial
and commercial areas in Gold Coast. This data set was obtained from a research
study which is currently being undertaken at QUT (Miguntanna 2009-unpublised
data). When comparing these sites with the sites investigated in this study, road
maintenance practices were assumed to be similar as they are maintained by the
same local government. Moreover, the data collection approach was also considered
to be similar for the three sites when compared to the sites which were investigated
in this research. Due to these reasons, use of this data set to check for near site
portability was considered appropriate.
The data set used is given in Appendix E. The data set includes all the parameters
measured in this research except turbidity. Therefore, the portability was not checked
for the relationships where turbidity is included (In Table 9.1, relationship number 5
and 8).
Using the surrogate parameter relationships which are given in Table 9.1, each key
parameter (DTN, TP, TDS and TS) was predicted using relevant parameter data from
the selected data set. For example, DTN was predicted using the relationship 1 in
Table 9.1, using TDS concentration data derived for the selected data set. The
predicted values for each parameter were then compared to the measured data in the
data set. The portability of the relationships derived for road surfaces are shown in
Figure 9.3a, 9.3b, 9.3c, 9.3d, 9.3e, 9.3f.
202
Figure 9.3a- Portability of the relationship 1- DTN-TDS relationship
0
2
4
6
0 2 4 6 DTN Observed
DTN
Pre
dict
ed
R2=0.98
Figure 9.3b -Portability of the relationship 2- DTN-DOC relationship
0
1
2
3
4
5
6
0 1 2 3 4 5 6DTN Observed
DT
N P
redi
cted
R2= 0.81
203
Figure 9.3c- Portability of the relationship 3- TP-TS relationship
Figure 9.3d- Portability of the relationship 4- TP-TOC relationship
0
1
2
3
4
0 1 2 3 4
TP Observed
TP
Pre
dict
ed
R2 =0.93
0
1
2
3
0 1 2 3
TP Obsevred
TP
Pre
dict
ed
R2= 0.50
204
Figure 9.3e- Portability of the relationship 6- TDS-EC relationship
Figure 9.3f- Portability of the relationship 7- TS-EC relationship
According to Figure 9.3a, 9.3b, 9.3c, 9.3d, 9.3e, 9.3f the following can be concluded:
• Relationship 1, relationship 2 and relationship 3 demonstrate good portability;
• The portability of relationship 4 is not recommended;
• Portability of relationship 6 and relationship 7 is poor.
0
100
200
300
400
500
0 100 200 300 400 500
TDS Obsevred
TD
S P
redi
cted
R2= 0.67
0
200
400
600
800
1000
1200
1400
1600
1800
0 200 400 600 800 1000 1200 1400 1600 1800
TS Observed
TS
Pre
dict
ed
R2= 0.65
205
Even though, relationship 1, relationship 2 and relationship 3 show good portability,
it is interesting to note that even for those relationships which demonstrated good
portability potential, some data substantially deviates from the 1:1 line. This may
probably be due to the atypical samples in each data set as discussed in Chapter 8
(Kayhanian et al. 2007; Thomson et al. 1997).
However, relationship 4, which is TP-TOC relationship does not show good
portability. This could be attributed to additional sources of phosphorus compounds
in the selected study sites. For example, the selected data set include data from a
industrial site which could have a high amount of TP generated from a concrete
batching plant in the area. Considering relationship 6 and 7, those relationships were
developed for TDS and TS by considering EC as the surrogate parameter. The reason
for poor portability of relationship 6 and 7 was not clear.
Where the relationships which are considered to be adequately portable, the
relationship coefficients (m) and coefficient of determination (R2) should be identical
to the m and R2 of relationships developed for the selected data set. However, in
practice exact portability of the relationships is unlikely (Thomson et al. 1997). This
can be discussed, in the context of changes in relationship coefficient (m) and
coefficient of determination (R2). For this purpose, the parameter relationships which
were identified as portable were developed for the selected data set separately. Table
9.3 summaries m and R2 for each relationship developed for the data set. In order to
compare the results, the m and R2 values for the relationships developed in this
research study was also included in Table 9.3.
As discussed above, relationship 1, relationship 2 and relationship 3 are noted as
portable relationships. However, the coefficients for these relationships show a slight
variation to each other (Table 9.3). These slight changes in relationship coefficients
could be due to a number of reasons. Firstly, the variation of degree of correlation
between the key parameter and its surrogate parameter may affect to the slight
difference of relationship coefficients obtained between sites. (Grayson et al. 1996).
Secondly, differences in wash-off behaviour can arise due to physical characteristics
such as traffic characteristics and surface characteristics which could also lead to
changes in the coefficients (Thomson et al. 1997). Therefore, it is possible that
206
relationship coefficients could be related to site-specific physical characteristics such
as drainage area, fraction of impervious area and average daily traffic which was not
investigated in this study. This would result in a more unique relationship for each
site and yet be a portable equation.
Table 9.3- Relationship coefficients (m) and coefficient of determination for the regression relationships
Relationship coefficient (m)
Coefficient of determination (R2)
Relationship1 DTN (mg/L) = m1TDS (mg/L)
The selected data set
0.006 0.81
This Study 0.013 0.82 Relationship 2 DTN (mg/L) = m2DOC (mg/L)
The selected data set
0.195 0.98
This Study 0.138 0.92 Relationship 3 TP (mg/L) = m3TS (mg/L)
The selected data set
0.003 0.93
This Study 0.002 0.84
9.4 Common surrogate parameter relationships for road and roof surfaces
Currently, most best management practices are focussed on providing treatment
measures directly for stormwater runoff at catchment outlets where separation of
road and roof runoff is not possible. Therefore, instead of a separate set of surrogate
parameter relationships for road runoff and roof runoff, identification of a common
set of parameters is more beneficial as these relationships can be used to evaluate
urban stormwater quality directly. According to the separate analysis of roads and
roofs which were carried out in this study, only DTN-TDS, TS-EC and TS-TTU was
identified as a common set of surrogates for both road and roof surfaces as discussed
in Section 9.2. Therefore, for only for these parameters, three relationships were
developed as given in Table 9.4 and Figure 9.4a, 9.4b, 9.4c. On the other hand, the
separate relationships derived for road and roof surfaces can be used individually,
where applicable.
207
Table 9.4- Common surrogate parameter relationships for road surfaces and roof surfaces
Key parameter
(Y)
Relationship Number
Surrogate parameter
(X)
Relationship Y = mX
Coefficient of determination
(R2)
Number of data points (N)
Standard error of
estimate to the mean Y SEE (%)
DTN 1 TDS DTN (mg/L) = 0.014TDS (mg/L) 0.86 76 45
2 EC TS (mg/L) = 1.450EC (mg/L) 0.83 50 44
TS
3 TTU TS (mg/L) = 14.369TTU (NTU) 0.88 72 40
208
y = 0.014xR2 = 0.86
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 100 200 300
TDS mg/L
DTN
mg/
L
Figure 9.4a- DTN-TDS relationship
y = 1.450xR2 = 0.83
0
20
40
60
80
100
120
140
160
0 50 100
TS
mg/
L
EC (µS/cm)
Figure 9.4b- TS-EC relationship
209
y = 14.369xR2 = 0.88
0
50
100
150
200
250
300
0 5 10 15 20
TTU (NTU)
TS
(m
g/L)
Figure 9.4c- TS-TTU relationship
9.5 Conclusions
Linear regression relationships were developed between key water quality
parameters and their surrogate parameters. Predictive ability of the relationships was
assessed based on the coefficient of determination (R2) and standard error of estimate
(SEE). Higher R2 with low SEE were considered as relationships with good
predictability.
For road surfaces, the surrogate parameter relationships were derived for nitrogen
compounds based on total dissolved solids and dissolved organic carbon. For
phosphorus compounds, relationships were derived for total phosphorus based on
total solids and total organic carbon. The study has also derived relationships for
solids based on turbidity and electrical conductivity.
For roof surface wash-off, a relationship was developed between dissolved total
nitrogen and total dissolved solids. Furthermore, relationships were derived for total
solids based on electrical conductivity and turbidity.
210
Near site portability of developed surrogate parameter relationships were only
assessed for the equations developed for road surface wash-off. A data set containing
pollutant wash-off data from industrial, residential and commercial sites in Gold
Coast was used to check the portability of the relationships developed. Relationships
obtained for DTN-TDS, DTN-DOC, TP-TS and TS-EC demonstrated good
portability potential. The portability of the relationship developed for TP and TOC
was found to be unsatisfactory. The relationship developed for TDS-EC and TS-EC
also demonstrated poor portability.
DTN-TDS, TS-EC and TS-TTU relationships were selected as common surrogate
parameter relationships for both road surfaces and roof surfaces.
211
Chapter 10 - Conclusions and Recommendations
10.1 Conclusions Water quality is typically expressed in terms of a range of water quality parameters.
Some of these parameters are easy to measure whilst others are difficult and
expensive to measure. This research study primarily focused on developing a set of
easy to measure water quality parameters which can be used as surrogate parameters
for other water quality parameters. Surrogate parameters were developed based on
the build-up and wash-off samples collected from selected urban surfaces. According
to research literature, build-up and wash-off are the two key pollutant processes that
define stormwater quality.
It was noted that road surfaces and roof surfaces represents the largest fraction of
impervious surfaces in an urban catchment. These two types of surfaces are the
primary contributors of pollutants to urban stormwater runoff. Therefore, build-up
and wash-off sampling was done on road and roof surfaces. In this regard, two road
surfaces namely, Drumbeat Street and Ceil Circuit in the Gold Coast were selected
as study sites. Additionally, two model roofs with different roofing materials were
used for pollutant build-up and wash-off investigations.
Selected road and roof surfaces represented the characteristics typical to the roads
and roofs in residential landuses. Samples were collected from small plot areas in
order to eliminate problems inherent in the use of non homogeneous areas. Build-up
sampling was conducted using a vacuuming cleaner. Wash-off investigations were
undertaken using simulated rainfall. This was to eliminate constraints inherent in the
use of natural rainfall events and its unpredictable occurrence. Wash-off samples
were collected using a specially designed vacuum system. All the collected samples
were tested for a range of physico-chemical water quality parameters.
It is understood that pollutant wash-off from urban impervious surfaces are
dependent on pollutant build-up. Therefore, prior to pollutant wash-off analysis,
212
pollutant build-up analysis was carried out. The main purpose of this analysis was to
gain a quantitative and qualitative understanding of pollutants present on road and
roof surfaces. Surrogate relationships were developed based on the physico-chemical
parameters tested for wash-off samples. For the development of surrogate
parameters, firstly, wash-off samples were analysed to identify a set of easy to
measure surrogate water quality parameters. Secondly, mathematical relationships
were developed from the selected surrogate parameters and the relevant parameters
of interest. Finally, relationships developed were validated using a separate data set
obtained from the Gold Coast area to assess the portability of the relationships
developed from this research.
10.1.1 Analysis of pollutant build-up
Analysis of pollutant build-up data from road surfaces revealed that total solids loads
obtained are typical to the road sites in the Gold Coast region (Egodawatta 2007;
Herngren et al. 2006). Total solids loads observed for road surfaces were 2595
mg/m2 and 962 mg/m2 for Drumbeat Street site and Ceil Circuit site respectively.
Differences in pollutant loads could be mainly due to the differences in the number
of antecedent dry days prior to sample collection. The samples collected from
Drumbeat Street belonged to 14 days of antecedent dry period whereas the
antecedent dry days for the samples collected at Ceil Circuit site was 7 days. This
further confirms that pollutant build-up on impervious surfaces varies considerably
with the antecedent dry period. The solids loads collected from roof surfaces were
around 180 mg/m2 which is typical to the amounts recovered from roof surfaces in
past research (Furumai et al. 2001; Van Metre and Mahler 2003). In comparison to
the road surfaces, the solids loads obtained from the roof surfaces were relatively
low. This could be attributed to the differences in surface characteristics such as
roughness, slope and different pollutant sources.
Particle size distribution analysis revealed that the solids build-up on road surfaces
was significantly finer. In Drumbeat Street site more than 92% of solids were finer
than 150 µm and in Ceil Circuit site around 77% solids were finer than 150 µm. This
agrees well with the findings of past researchers who noted that the particle size
213
distribution of pollutant build-up on Australian road surfaces is significantly finer. In
regards to roof surfaces, it was noted that around 80% of solids were finer than 150
µm. Therefore, it can be surmised that roof surfaces also contains a significant
amount of fine solids similar to the road surfaces.
Investigations into the physico-chemical characteristics of pollutant build-up resulted
in understanding the nature of the pollutants on each impervious surface. Solids on
road surfaces contained a higher loading of organic matter than nutrients. This
indicates that the build-up on road surfaces is organically rich. On the other hand,
unlike road surfaces, similar to the contribution of organic carbon load, total nitrogen
and total phosphorus also contribute considerable loads to total solids load on roof
surfaces.
A relatively higher amount of nitrogen and phosphorus compounds were found to be
in the particle size ranges <150 µm which confirms the highly polluted nature of the
finer fraction of pollutant build-up. The analysis of pollutant build-up data for
different particle size fractions of solids from both road and roof surfaces clearly
confirmed the highly polluted nature of road surface build-up in comparison to the
roof surfaces. Hence a separate analysis of roads and roofs was undertaken in wash-
off analysis.
10.1.2 Analysis of pollutant wash-off
Prior to the analysis of physico-chemical parameters to identify suitable surrogate
parameters, wash-off data was evaluated to check the appropriateness of the use of
simulated rainfall to generate wash-off samples. This was done by comparing the
variations of solids wash-off process observed in this research with the general
understanding of the wash-off process.
In this regard, the variation of total solids concentrations with rainfall intensity and
duration and the variation of particle size distribution were analysed. It was found
that the concentration of total solids in wash-off was higher for the shorter durations
for each intensity compared to the longer durations. The concentration decreased
214
exponentially with the increase in duration for all the intensities. Analysis of particle
size distribution revealed that the wash-off of the fine fraction of solids (<150 µm)
increased with the increase in the rainfall intensity. This confirmed that the
variations in solids wash-off observed in this research agrees with the general
understanding of the wash-off process in quantitative and qualitative terms.
Identification of surrogate parameters was done in two steps. Firstly, surrogate
parameters were identified for key water quality parameters of interest. This was
done for road and roof surfaces separately. This separation was due to the
significantly higher pollutant concentrations noted in wash-off from road surfaces
compared to roof surfaces. Secondly, a common set of surrogate parameter
relationships were identified for road and roof surfaces. Multivariate data analysis
techniques, namely, Principal component analysis (PCA) and Partial least squares
regression (PLS) were used to identify the surrogate parameters and to develop
relationships between the selected surrogate parameters and key water quality
parameters of interest.
Table 10.1 and Table 10.2 shows the identified surrogate parameter relationships for
both road and roof surfaces. Among the parameters measured in this research, EC,
TTU, TSS, TDS, TOC and DOC were considered as the easiest to measure a set of
parameters. Therefore, special attention was given to finding correlations for
nitrogen and phosphorus compounds with EC, TTU, TSS, TDS, TS, TOC and DOC.
It was determined that dissolved total nitrogen (DTN) and total phosphorus (TP) are
the most representative parameters from all the nitrogen and phosphorus compounds
in urban stormwater runoff. Therefore, relationships were developed only for DTN
and TP. The surrogate parameters identified for DTN and TP would be suitable for
all other nitrogen and phosphorus compounds. Additionally, surrogate parameter
relationships were also developed for TSS, TDS and TS as these are the key
indicators of solids in urban stormwater runoff.
215
Table 10.1- Surrogate parameter relationships for road surfaces
Key parameter
Surrogate parameter (X)
Relationship Y = mX
TDS DTN (mg/L) = 0.013TDS (mg/L) DTN
DOC DTN (mg/L) = 0.138DOC (mg/L)
TS TP (mg/L) = 0.002TS (mg/L) TP
TOC TP (mg/L) = 0.020TOC (mg/L)
TSS TTU TSS (mg/L) = 1.982TTU (NTU)
TDS EC TDS (mg/L) = 2.195EC (µS/cm)
EC TS (mg/L) = 2.735EC (µS/cm) TS
TTU TS (mg/L) = 14.281TTU (NTU)
Table 10.2- Surrogate parameter relationships for roof surfaces Key parameter
(Y) Surrogate parameter
(X)
Relationship Y = mX
DTN TDS DTN (mg/L) = 0.011TDS (mg/L)
EC TS (mg/L) = 0.759EC (µS/cm) TS
TTU TS (mg/L) = 10.640TTU (NTU)
Portability of the developed relationships was evaluated using an available data set.
Portability refers to the degree to which the developed surrogate parameter
relationships are applicable for the prediction of key water quality parameters in a
geographical area different to the area used for the derivation of the relationships.
Consequently, near site portability of the developed surrogate parameter
relationships was assessed only for road surface wash-off as a separate data set for
roof surfaces could not be found. The analysis revealed the relationships obtained for
DTN-TDS, DTN-DOC and TP-TS demonstrated good portability potential. The
portability of the relationship developed for TP and TOC was not satisfactory. The
relationship developed for TDS-EC and TS-EC also demonstrated poor portability.
216
Possibility of developing common surrogate relationships for both road and roof
surfaces was also assessed. This was done due to the inherent difficulties in
separately analysing road and roof runoff contributions from urban catchments.
Relationships shown in Table 10.3 are common relationships for both road and roof
surfaces. Only DTN-TDS, TS-EC and TS-TTU was found to be a common set of
surrogates for both road and roof surfaces. No common relationships were obtained
for TP, TSS and TDS. However, the separate relationships derived for road and roof
surfaces can be used individually, where applicable.
Table 10.3- Common Surrogate parameter relationships for road surfaces and roof surfaces Key parameter
(Y) Surrogate parameter
(X) Relationship
Y = mX
DTN TDS DTN (mg/L) = 0.014TDS (mg/L)
EC TS (mg/L) = 1.450EC (mg/L) TS
TTU TS (mg/L) = 14.369TTU (NTU)
10.2 Recommendations for further research This research developed a set of surrogate water quality parameters to evaluate urban
stormwater quality. The identified surrogate parameter relationships provide an easy
approach to measure selected key water quality parameters. These relationships have
the potential to enhance the acquisition of important information on urban
stormwater quality without resource intensive laboratory based measurements of key
water quality parameters. Furthermore, the findings of this research strengthens the
current knowledge on pollutant build-up and wash-off processes on road and roof
surfaces.
A number of knowledge gaps that limit the general understanding of stormwater
quality were identified during this study. Following are the recommendations for
future research to overcome current knowledge gaps:
217
• Validation of the relationships developed in this study using natural rainfall
data is recommended. This will enhance the transferability of outcomes
generated by this study.
• Relationship coefficients could be predictable using site specific physical
characteristics such as drainage area, percentage of impervious area and annual
daily traffic. This would result in a unique relationship for each site, yet be a
portable relationship. Therefore, it is recommended to investigate possible
relationships between the relationships coefficients and the physical
characteristics of each site.
• In this study, only near site portability for road surfaces was tested. It is
recommended to assess the portability of the relationships developed for far
sites with different catchment characteristics and management practices. This
will improve the applicability of the developed relationships.
• The relationships developed especially for roof surfaces have been based on a
relatively small data set. Further monitoring is therefore recommended to
improve the reliability of the relationships.
• This study specifically focussed on nitrogen and phosphorus compounds as key
pollutants. Further study is needed to develop the surrogate parameter
relationships among other key pollutant indicators including hydrocarbons and
heavy metals.
• In this study, the number of build-up samples collected was limited. The
limited number of samples for pollutant build-up investigations constrained the
applicability of multivariate statistical methods for data analysis. It is
recommended that additional build-up samples should be collected to apply
further rigorous multivariate data analysis. In this context, the collection of
build-up samples by considering changes in antecedent dry days is also
recommended.
• The understanding gained from correlations among physico-chemical
parameters and the identified surrogate parameter relationships were limited for
bulk wash-off samples. However, it is important to assess the appropriateness
of the identified correlations among the parameters for fractionated wash-off
samples where the total solids load has been separated into different particle
size ranges. This is because some monitoring programs may require the
218
analysis of physico-chemical parameters in different particle size ranges of
wash-off solids.
219
References 1. Adams, M.J. (1995). Chemometrics in analytical spectroscopy. Cambridge
[England], royal Society of Chemistry.
2. Aguilera, P.A., Frenich, A.G., Castro, H., Martinez, J. L. (2000). PLS and PCR
Methods in the assessment of coastal water quality. Environmental Monitoring
and Assessment, Vol. 62, pp. 193-204.
3. AIM600 User Manual (2005), AIM600 Block Digestion System- User Manual.
4. Alberto, W.D., Del Pilar, D.M., Valeria, A.M., Fabiana, P.S., Cecilia, H.A. and
Angeles, B.M.D.L. (2001). Pattern recognition techniques for the evaluation of
spatial and temporal variations in water quality. A case study: Suquia river
basin (Cordoba-Argentina).Water Research, Vol. 35 (No. 12), pp. 2881-2894.
5. Allison, R.A., Chiew, F.H.S. and McMahon, T.A. (1998). Nutrient contribution
of leaf litter in urban stormwater. Journal of Environmental Management, Vol.
54, pp. 269-272.
6. Andral, M.C., Roger, S., Montrejaud-Vignoles, M. and Herremans, L. (1999).
Particle size distribution and hydrodynamic characteristics of solid matter
carried by runoff from motorways. Water Environment Research, Vol. 71 (No.
4), pp. 398-407.
7. Anon (1981). Characterization of pollution in urban storm water runoff.
Australian Water Resources Council Technical Paper No 60.
8. APHA, (2005). Standard Methods for the Examination of Water and Waste
Water, Washington, D.C: American Public Health Association.
220
9. AQ 2 Discrete Analyser Operator Manual (2006). Seal Analytical Operator
Hand Book.
10. Arcy, B.D. and Frost, A. (2001). The role of best management practices in
alleviating water quality problems associated with diffuse pollution. The
Science of the Total Environment, Vol. 265, pp. 359-367.
11. Arnold, C.L.Jr. and Gibbons, C.J. (1996). Impervious surface coverage:
Emergence of a key environmental factor. Journal of the American Planning
Association, Vol. 62 (2), pp. 243-258.
12. Arntson, A.D. and Tornes, L.H. (1985). Rainfall-runoff relationships and
water-quality assessment of Coon Creek watershed, Anoka County, Minnesota.
Water Resources and Management.
13. AS/NZS, (1998). 5667.1- Australian New Zealand Standards, Water quality-
Sampling, Unpublished manuscript.
14. Assouline, S., El Idrissi, A. and Persoons, E. (1997). Modelling the physical
characteristics of simulated rainfall: a comparison with natural rainfall. Journal
of Hydrology, Vol. 196 (No. 1-4), pp. 336-347.
15. Atasoy, M., Palmquist, B. and Phaneuf, D.J. (2006). Estimating the effects of
urban residential development on water quality using microdata. Journal of
Environmental Management, Vol. 79, pp. 399-408.
16. Atekwanaa, E.A., Atekwanaa, E.A. and Roweb, R.S. (2004). The relationship
of total dissolved solids measurements to bulk electrical conductivity in an
aquifer contaminated with hydrocarbon. Journal of Applied Geophysics, Vol.
56, pp. 281- 294.
17. Ayoko, G.A., Singh, K., Balera, S. and Kokot, S. (2007). Exploratory
multivariate modelling and prediction of the physico- chemical properties of
surface water and groundwater. Journal of Hydrology, Vol. 336, pp. 115-124.
221
18. Badin, A., Faure, P., Bedell, J. and Delolme, C. (2008). Distribution of organic
pollutants and natural organic matter in urban stormwater sediments as a
function of grain size. Science of the Total Environment, Vol. 403, pp. 178-
187.
19. Bahar, Md. M., Ohmori, H. and Yamamuro, M. (2007). Relationship between
river water quality and land use in a small river basin running through the
urbanising area of Central Japan. The Japanese Society of Limnology, Vol. 9,
pp. 19-26.
20. Baird, C. (1999a). Toxic organic chemicals. Environmental chemistry, W.H.
Freeman and Company, pp. 381-418.
21. Balades, J.D., Cathelain, M., Marchandise, P., Peybernard, J. and Pilloy, J.C.
(1984). Chronic pollution of intercity motorway runoff waters. Water Science
Technology, Vol. 17 (No. 6-7), pp. 1165-1174.
22. Ball, J.E., Wojcik, A. and Tilley, J. (2000). Stormwater quality from road
surfaces-monitoring of the Hume highway at south Strathfield. Research report.
23. Ball, J.E., Jenks, R. and Aubourg, D. (1998). An assessment of the availability
of pollutant constituents on road surfaces. The Science of the Total
Environment Vol. 209, pp. 243-254.
24. Bannerman, R.T., Owens, D.W., Dodds, R.B. and Hornewer, N.J. (1993).
Sources of pollutants in Wisconsin stormwater. Water Science and
Technology, Vol. 28 (No. 3-5), pp. 241-259.
25. Bannerman, R.T., Baun, K., Bohn, M., Huges, P.E. and Graczyk, D.A. (1983).
Evaluation of urban non-point source pollution management in Milwaukee
country, Wisconsin, Vol. 1, EPA, Water planning division.
222
26. Barbara, M.S., Ahif, S., Felipe, V. and Rullkotter, J. (2009). Natural vs.
anthropogenic sources of hydrocarbons as revealed through biomarker
analysis: A case study in the southern Gulf of Mexico.
27. Barnes, K.B., Morgan, J.M. and Martin C. (2001). Impervious surfaces and the
quality of natural and built environments, Roberge Department of Geography
and Environmental Planning, Towson University, Baltimore, Maryland.
28. Barrett, M.E., Irish Jr. L.B., Malina, Jr. J.F. and Charbeneau, R.J. (1998).
Characterisation of highway runoff in Austin, Texas, area. Journal of
Environmental Engineering, Vol. 124 (No. 2), pp. 131-137.
29. Barrett, M.E., Zuber, R.D., Malina, J.F., Charbeneau, R. and Ward, G.H.
(1995). A review and evaluation of literature pertaining to the quantity and
control of pollution from highway runoff and construction. CRWR Online
Report 95-5, Centre for research in water resources, The University of Texas,
Austin.
30. Barrios, A. (2000). Urbanisation and Water Quality. CAE Working paper
series, WP00-1. American Farmland Trust’s Centre for Agriculture in the
Environment, DeKalb.
31. Batten, G.D. (1998). Plant analysis using near infrared reflectance
spectroscopy: The potential and the limitations. Australian Journal of
Experimental Agriculture, Vol. 38, pp. 697–706.
32. Beckwith, P.R., Warren, R.S. and Harrop, D.O. (1984). The behaviour of
sediment associated heavy metals within an urban surface runoff drainage
system. Urban Pollution Resource Centre, UK, pp. 785-789.
33. Berdahl, P., Akbari, H., Levinson, R. and Miller, W.A. (2008). Weathering of
roofing materials-An overview. Construction and Building Materials, Vol. 22,
pp. 423-433.
223
34. Berretta, C., Gnecco, I., Lanza, L.G. and La Barbera, P. (2007). Hydrologic
influence on stormwater pollution at two urban monitoring sites. Urban Water
Journal, Vol. 4 (No. 2), pp. 107-117.
35. Bian, B. and Zhu, W. (2008). Particle size distribution and pollutants in road-
deposited sediments in different areas of Zhenjiang, China. Environmental
Geochem Health.
36. Brabec, E., Schulte, S. and Richards, P.L. (2002). Impervious surfaces and
water quality: A review of current literature and its implications for watershed
planning. Journal of Planning Literature, Vol. 16 (No. 4), pp. 499-514.
37. Bradford, W.L. (1977). Urban stormwater pollutant loadings: a statistical
summary. Journal of Water Pollutant Control Federation, Vol. 49, pp. 613-622.
38. Brans, J.P. (2002). Ethics and decision. European Journal of Operational
Research, Vol. 136, pp. 340-352.
39. Brans, J.P., Vincke, Ph. and Mareschal, B. (1986). How to select and how to
rank projects, The PROMETHEE method. European Journal of Operational
Research, Vol. 24, pp. 228-238.
40. Brezonik, P.L. and Stadelmann, T.H. (2002). Analysis and predictive models of
stormwater runoff volumes, loads and pollutant concentrations from
watersheds in the twin cities Metropolitan Area, Minnesota, USA. Journal of
Water Research, Vol. 36, pp. 1743-1757.
41. Brinkmann, W.L.F. (1985). Urban stormwater pollutants: Sources and
loadings. Geo Journal, Vol. 11 (No. 3), pp. 277-283.
42. Bris, F.J., Garnaud, S., Apperry, N., Gonzalez, A., Mouchel, J.M. and Chebbo,
G., Thevenot, D.R. (1999). A Street deposit sampling method for metal and
hydrocarbon contamination assessment. The Science of the Total Environment,
Vol. 121, pp. 229–237.
224
43. Brodie, I.M. (2007). Investigation of stormwater particles generated from
common urban surfaces. Thesis- University of Southern Queensland, Australia.
44. Bubenzer, G.D., Molnau, M. and McCool, D.K. (1984). Low intensity rainfall
with a rotating disk simulator. Transactions of the ASAE, pp. 1230-1232.
45. Bujon, G., Herremans, L. and Phan, L. (1992). A forecasting model for flow
and pollutant discharge from sewerage systems during rainfall events. Water
Science and Technology, Vol. 25 (No. 8), pp. 207-215.
46. Button, G.A. and Pitt, E.B. (2001). Stormwater effects Handbook. Lewis
Publishers.
47. Campbell, M.R., Brumm, T.J. and Glover, D.V. (1997). Whole grain amylase
analysis in maize using near- infrared transmittance spectroscopy. Cereal
Chemistry, Vol. 74 (No. 3), pp. 300-303.
48. Carmody, O., Kristof, J., Frost, R.L., Mako, E., Kloprogge, J.T. and Kokot, S.
(2005). A spectroscopic study of mechanochemically activated kaolinite with
the aid of chemometrics, Journal of Colloid and Interface Science, Vol. 287,
pp. 43-56.
49. Carroll, S. and Goonetilleke, A. (2005). Assessment of high density of onsite
wastewater treatment systems on a shallow groundwater coastal aquifer using
PCA. Environmetrics, Vol. 16, pp. 257-274.
50. Cave, K.A., and L.A. Roesner. (1994). Overview of stormwater monitoring
needs. Proceedings of the Engineering Foundation Conference on Stormwater
Monitoring, August 7-12, Crested Butte, CO.
51. Chang, C.W., Laird, D.A., Mausbach, M.J. and Hurburgh, C.R. (2001). Near-
Infrared reflectance spectroscopy- Principal components regression analyses of
soil properties. Soil Science Vol. 65, pp. 480-490.
225
52. Chang, M., McBroom, M.W. and Beasley, R.S. (2004). Roofing as a source of
non-point water pollution. Journal of Environmental Management, Vol. 73, pp.
307-315.
53. Chang, M. and Crowley, C.M. (1993). Preliminary observations on water
quality of storm runoff from four selected residential roofs. Water Resources
Bulletin, American water resources association, Vol. 29 (No. 5), pp. 777-783.
54. Chang, Y., Chou, C., Su, K. and Tseng, C. (2005). Effectiveness of street
sweeping and washing for controlling ambient TSP. Atmospheric
Environment, Vol. 39, pp. 1891-1902.
55. Chapman, D. (1992). Water quality assessments. A guide to the use of biota,
sediments and water in environmental monitoring.
56. Charlesworth, S.M. and Lees, J.A. (1999). The distribution of heavy metals in
deposited urban dusts and sediments, Coventry, England. Environmental
Geochemistry and Health, Vol. 21, pp. 97-115.
57. Chebbo, G. and Gromaire, M.C. (2004). The experimental urban catchment -Le
Marais in Paris: what lessons can be learned from it? Journal of Hydrology,
Vol. 299, pp. 312-323.
58. Chen, J. and Barry, J.A. (2006). Analytical urban stormwater quality models
based on pollutant build-up and wash- off processes. Journal of Environmental
Engineering, Vol. 132 (No. 10), pp. 1314-1330.
59. Chernova, T.G., Rao, P.S., Pikovskii, Y.I., Alekseeva, T.A., Nath, B.N., Rao,
B.R. and Rao, Ch.M. (2001). The composition and the source of hydrocarbons
in sediments taken from the tectonically active Andaman Backarc basin, Indian
ocean. Marine Chemistry, pp. 1-15.
226
60. Chiew, F.H.S. and McMahon, T.A. (1999). Modelling runoff and diffuse
pollution loads in urban areas. Water Science and Technology, Vol. 39 (No.
12), pp. 241-248.
61. Christopherson, R. W. (2001). Elemental Geosystems. 3rd edition. Upper
Saddle River, NJ:Prentice-Hall.
62. Christiansen, J. P. (1942). Irrigation by sprinkling. Bulletin No 670, University
of California Agricultural Experiment Station.
63. Chui, P.C. (1997). Characteristics of stormwater quality from two urban
catchments in Singapore. Environmental Monitoring and Assessment, Vol. 44,
pp. 173-181.
64. Chrystal, C.P. (2006). The Characterisation of stormwater runoff from road
surfaces. Thesis- University of KwaZulu- Natal, Durban, South Africa.
65. Connecticut stormwater manual (2004).Why stormwater matters: the impacts
of urbanisation. The Connecticut Department of Environmental Protection.
66. Connel, D. (1993). Metals, minerals and suspended solids. Water Pollution,
University of Queensland Press, St.Lucia, Queensland, Australia, pp. 118-143.
67. Coomera Waters Ltd. (2009). Coomera Waters Ltd web page-
www.coomerawaters.com.au
68. Cordery, I. (1977). Quality characteristics of urban stormwater in Sydney,
Australia. Water Resources Research, Vol. 13 (No. 1), pp. 197-201.
69. Cordery, I. (1976). Some effects of urbanisation on streams. Civil Engineering
Transactions, Vol. CE18, Institution of Engineers Australia, pp. 7-11.
227
70. Datry, T., Malard, F., Vitry, L., Hervant, F. and Gibert, J. (2003). Solute
dynamics in the bed sediments of a stormwater infiltration basin. Journal of
Hydrology Vol. 273, pp. 218-233.
71. Deb, D., Deshpande, V.N. and Das, K.C. (2008). Assessment of water quality
around surface coal mines using principal component analysis and fuzzy
reasoning techniques. Mine Water Environment, Vol. 27, pp. 183-193.
72. Decision Lab 2000, Executive Edition, Getting Started Guide, Visual Decision
Inc. 2000.
73. Deletic, A and Orr, D.W. (2005). Pollution build-up on road surfaces. Journal
of Environmental Engineering, Vol. 131 (No. 1), pp. 49-59.
74. Deletic, A. (1998). The first flush load of urban surface runoff. Water
Resources, Vol. 32 (No. 8), pp. 2462-2470.
75. Deletic, A.B. and Maksimovic, C.T. (1998). Evaluation of water quality factors
in storm runoff from paved areas. Journal of Environmental Engineering, Vol.
124 (No. 9), pp. 869-879.
76. De Lima, J.L.M.P. and Singh, V.P. (2003). Laboratory experiments on the
influence of storm movement on overland flow. Physics and Chemistry of the
Earth, Vol. 28, pp. 277-282.
77. Dong, A., Chesters, G., and Simsiman, G.V. (1984). Metal composition of soil,
sediments and urban dust and dirt samples from the Menomonee river
watershed, Wisconsin, USA. Water, Air and Soil Pollution, Vol. 22, pp. 257-
275.
78. DR/4000, Spectrophotometer instrument manual (1999).
228
79. Duncan, H.P. (1995). A review of urban stormwater quality processes. Report
No 95/9, Cooperative Research Centre for Catchment Hydrology, Melbourne,
Australia.
80. Duncan, M., Jain, R., Yung, S.C. and Patterson, R. (1985). Performance
evaluation of an improved street sweeper. US Environmental Protection
Agency (US EPA-600/7-85-008).
81. Dunn, B.W., Beecher, H.G., Batten, G.D. and Ciavarella, S. (2002). The
potential of near-infrared reflectance spectroscopy for soil analysis- A case
study from the riverine plain of south-eastern Australia. Australian journal of
Experimental Agriculture, Vol. 42, pp. 607-614.
82. Egodawatta, P. (2007). Translation of small plot pollution mobilisation and
transport measurements from impermeable urban surfaces to urban catchment
scale. Thesis-Queensland University of Technology, Brisbane Australia.
83. Egodawatta, P. and Goonetilleke, A. (2008). Modelling pollutant build-up and
wash-off in urban road and roof surfaces. Proceedings of the 31st Hydrology
and Water Resources Symposium and the 4th International Conference on
Water Resources and Environment Research, Engineers Australia, Adelaide,
Australia.
84. Egodawatta, P., Thomas, E., Goonetilleke, A. (2007). Mathematical
interpretation of pollutant wash-off from urban road surfaces using simulated
rainfall. Water Research, pp. 3025-3031.
85. Egodawatta, P., Goonetilleke, A., Ayoko, G.A. and Thomas, E. (2006).
Understanding the interrelationship between stormwater quality and rainfall
and runoff factors in residential catchments. Proceedings of the 7th
International Conference on Urban Drainage Modelling and the 4th
International Conference on Water Sensitive Urban Design, Melbourne,
Australia, pp. 43-50.
229
86. Egodawatta, P. and Goonetilleke, A. (2006). Characteristics of pollutants built-
up on residential road surfaces. Proceedings of the 7th International Conference
on HydroScience and Engineering Philadelphia, USA September 10-13.
87. Einax, J.W., Truckenbrodt, D. and Kampe, O. (1998). River pollution data
interpreted by means of chemometric methods. Microchemical Journal, Vol 58,
pp. 315-324.
88. Ellis, J.B. (1991). Urban runoff quality in the UK: Problems, prospects and
procedures. Applied Geography, Vol. 11, pp. 187-200.
89. Ellis, J.B. (1986). Pollution aspects of urban runoff. Urban runoff pollution,
Vol. G10, pp. 1-31.
90. Ellis, J.B., Hamilton, R. and Roberts, A.H. (1981). Composition of suspended
solids in urban stormwater. Urban Stormwater Quality, Management and
Planning, Proceedings of the Second International Conference on urban storm
drainage, Water Resources Publications, Colorado, pp. 184-190.
91. Ellis, J.B., Revitta, D.O., Harrop, A. and Beckwitha, P.R. (1987). The
contribution of highway surfaces to urban stormwater sediments and metal
loadings. Urban Pollution Research Centre, Middlesex Polytechnic,
Queensway, Enfield, Middlesex EN3 4SF, UK.
92. Eric, W. and Strecker, P.E. (2000). Considerations and Approaches for
Monitoring the Effectiveness of Urban BMPs, Woodward-Clyde, Portland,
Oregon.
93. Eriksson, L., Hagberg, P., Johansson, E., Rannar, S., Whelehan, O., Astrom, A.
and Lindgren, T. (2001). Multivariate process monitoring of a newsprint mill.
Application to modelling and predicting COD load resulting from de-inking of
recycled paper. Journal of Chemometrics, Vol. 15, pp. 337-352.
230
94. Espey, W.H., Winslow, D.E. and Morgan, C.W. (1969). Urban effects on the
unit hydrograph. Proceedings of the Second Water Resources Symposium, pp.
215-228.
95. Espinasse, B., Picolet, G. and Chouraqui, E. (1997). Negotiation support
systems: A multi-criteria and multi-agent approach. European Journal of
Operational Research, Vol. 103, pp. 389-409.
96. Eyre, B.D. and Pepperell, P. (1999). A spatially intensive approach to water
quality monitoring in the Rous river catchment, NSW, Australia. Journal of
Environmental Management, Vol. 56, pp. 97-118.
97. Faber, K. and Kowalski, B. (1997). Propagation of measurement errors for the
validation of predictions obtained by principal component regression and
partial least squares. Journal of Chemometrics, Vol. 11, pp. 181-238.
98. Falco, P.C., Genaro, L.A.T., Lloret S.M. and Reig, F.B. (2002). Multivariate
calibration applied to simultaneous chemiluminescence determination of cobolt
and chromium. Analytical and bioanalytical chemistry, Vol. 374, pp. 1223-
1229.
99. Farahmand, T., Fleming, S.W. and Edward, J. (2006). Detection and
visualization of storm hydrograph changes under urbanization: An impulse
response approach. Journal of Environmental Management.
100. Farnham, I.M., Johannesson, K.H., Singh, A.K., Hodege, V.F. and
Stetzenbach, K.J. (2003). Factor analytical approaches for evaluating
groundwater trace element chemistry data. Analytica Chimica Acta, Vol. 490,
pp.123-138.
101. Ferrer, R., Beltran, J.L. and Guiteras, J. (1998). Multivariate calibration applied
to synchronous fluorescence spectrometry. Simultaneous determination of
polycyclic aromatic hydrocarbons in water samples. Talanta, Vol. 45, pp. 1073-
1080.
231
102. Flanagan, D.C. and Foster, G. R. (1989). Storm pattern effect on nitrogen and
phosphorus losses in surface runoff. Water Resources and Management, Vol.
32 (No. 2), pp. 535-544.
103. Fletcher, T., Duncan, H., Poelsma, P. and Lioyd, S. (2004). Storm water flow
and quality and the effectiveness of non-proprietary stormwater treatment
measures-A review and gap analysis, Technical report, Report 04/8, December,
Cooperative research centre for catchment hydrology.
104. Forster, J. (1996). Patterns of roof runoff contamination and their potential
implications on practice and regulation of treatment and local infiltration.
Water Science and Technology, Vol. 33, pp. 39-49.
105. Forster, J. (1999). Variability of roof runoff quality. Water Science and
Technology, Vol. 39 (No 55), pp. 137-144.
106. Furumai, H., Hijioka, Y. and Nakajima, F. (2001). Modelling and field survey
of wash-off behaviour of suspended particles from roofs and roads. Conference
Proceeding Paper, American Society of Civil Engineers.
107. Gadd, J. and Kennedy, P. (2001). House roof runoff: Is it clean as we think.
Paper presented in 2nd South Pacific stormwater conference.
108. Geiger, W.F. (1987). Flushing effects in combined sewer systems. In Urban
stormwater quality: planning and management. Ecole polytechnique Federale,
IAHR/IAWPRC.
109. Gippel, C.J. (1995). Potential of turbidity monitoring for measuring the
transport of suspended solids in streams. Hydrological Processes, Vol. 9, pp.
83-97.
110. Gnecco, I., Berretta, C., Lanza, L.G. and Barbera, P.L. (2005). Stormwater
pollution in the urban environment of Genoa, Italy. Atmospheric Research,
Vol. 77, pp. 61-73.
232
111. Gobel, P., Dierkes, C. and Coldewey, W.G. (2006). Stormwater runoff
concentration matrix for urban areas. Journal of Contaminant Hydrology, In
press, Corrected Proof.
112. Google Map data (2009). Google maps web page-www.maps.google.com.au
113. Goonetilleke, A., Egodawatta, P. and Kitchen, B. (2009). Evaluation of
pollutant build-up and wash-off from selected land uses at the Port of Brisbane,
Australia. Marine pollution bulletin, Vol. 58, pp. 213-221.
114. Goonetilleke, A., Thomas, E., Ginn, S., Gilbert, D. (2005). Understanding the
role of land use in urban stormwater quality management. Journal of
Environmental Management Vol. 74, pp. 31-42.
115. Goonetilleke, A. and Thomas, E. (2004). Water quality impacts of
urbanisation, relating water quality to urban form. Research Report, Energy
and resource management research program, Centre for built Environment and
Engineering, Queensland university of technology, Australia.
116. Grayson, R.B., Finlayson, B.L., Gippel, C.J. and Hart, B.T. (1996). The
potential of field turbidity measurements for the computation of total
phosphorus and suspended solids loads. Journal of Environmental
Management, Vol. 47, pp. 257-267.
117. Grayson, R.B., Gippel, C.J., Finlayson, B.L. and Hart, B.T. (1997). Catchment-
wide impacts on water quality: the use of 'snapshot' sampling during stable
flow. Journal of Hydrology, Vol. 199, pp. 121-134.
118. Gromaire-Mertz, M.C., Garnaud, S., Gonzalez, A. and Chebbo, G. (1999).
Characterisation of urban runoff pollution in Paris. Water Science and
Technology, Vol. 39 (No. 2), pp. 1-8.
119. Gulson, B.L., Davis, J.J., Mizona, K.J., Korscha, M.J. and Bawden-Smithb, J.
(1995). Sources of lead in soil and dust and the use of dust fallout as a
233
sampling medium. The Science of the Total Environment, Vol. 166, pp. 245-
262.
120. Guo, H., Wang, T. and Louie, P.K.K. (2004). Source apportionment of ambient
non- methane hydrocarbons in Hong Kong: Application of a principal
component analysis/absolute principal component scores (PCA/ APCS)
receptor model. Environmental Pollution, Vol. 129, pp. 489-498.
121. Hall, M.J. and Ellis, J.B. (1985). Water quality problems of urban areas. Geo
Journal, Vol. 11 (No. 3), pp. 265-275.
122. Hallberg, M. (2006). Suspended solids and metals in highway runoff-
implications for treatment systems. TRITA LWR LIC Thesis 2035.
123. Han, Y., Lau, S., Kayhanian, M. and Stenstorm, M.K. (2006). Characteristics
of highway stormwater runoff. Water Environment Research, Vol. 78 (No. 12),
pp. 2377-2388.
124. Harmel, R.D., King, K.W., Haggard, B.E., Wren, D.G. and Sheridan, J.M.
(2006). Practical guidance for discharge and water quality data collection on
small watersheds. American Society of Agricultural and Biological Engineers,
Vol. 49 (No. 4).
125. Harned, D.A. (1988). Effects of highway runoff on streamflow and water
quality in the Sevenmile Creek basin, a rural area in the Piedmont Province of
north Carolia, United States Geological Survey, Water supply paper 2329.
126. Herngren, L. (2005). Build-up and wash-off process kinetics of PAHs and
heavy metals on paved surfaces using simulated rainfall. Thesis- Queensland
University of Technology, Brisbane, Australia.
127. Herngren, L., Goonetilleke, A. and Ayoko, G.A. (2005). Understanding heavy
metal and suspended solids relationships in urban stormwater using simulated
rainfall. Journal of Environmental Management, Vol. 76, pp. 149-158.
234
128. Herngren, L., Goonetilleke, A. and Ayoko, G.A. (2006). Analysis of heavy
metals in road- deposited sediments. Analytica Chimica Acta, Vol. 571, pp.
270-278.
129. Hijioka, Y., Nakajima, F. and Furumai, H. (2001). Modified models of wash-
off from roofs and roads for non-point pollution analysis during first flush
phenomena. Conference Proceeding Paper, American Society of Civil
Engineers.
130. Hoffman, E.J., Latimer, J.S., Hunt, C.D., Mills, G.L. and Quinn, J.G. (1984).
Stormwater runoff from highways. Water, Air and Soil Pollution, Vol. 25, pp.
349-364.
131. Hoffman, E.J., Latimer, J.S., Hunt, C.D., Mills, G.L. and Quinn, J.G. (1985).
Stormwater runoff from highways. Water, Air and Soil Pollution, Vol. 25, pp.
349-364.
132. Hope, D., Naegeli, M.W., Chan, A.H. and Grimm, N.B. (2004). Nutrients on
asphalt parking surfaces in an urban environment. Water, Air and Soil
Pollution, Vol. 4, pp. 371-390.
133. Horner, R.R., Burges, S.J., Ferguson, J.F., Mar, B.W. and Welch, E.B. (1979).
Highway runoff monitoring: The initial year, WA-RD-39.3. Washington State
department of transportation, Olympia, W.A.
134. House, M.A., Ellis, J.B., Herricks, E.E., Hvitved-Jacobsen, T., Seagar, J.,
Lijklema, L., Aalderink, H. and Clifforde, I.T. (1993). Urban drainage-impacts
on receiving water quality. Water Science and Technology, Vol. 27 (No. 12),
pp. 117-158.
135. Howell R.B. (1978). Water pollution aspects of particles which collect on
highway surfaces. FHWA-CA-78-22, California Department of Transportation,
Sacramento, C.A.
235
136. Huang, J., Du, P., Ao, C., Ho, M., Lei, M., Zhao, D. and Wang, Z. (2007).
Multivariate Analysis for Stormwater Quality Characteristics Identification
from Different Urban Surface Types in Macau. Bullatin Environ. Contam
Toxicol, Vol. 79, pp. 650-654.
137. Hudson, N.W. (1963). Raindrop size distribution in high intensity storms.
Rhodesian Journal of Agricultural Research, Vol. 1, 6-11.
138. Hunter, J.V., Sabatino, T., Gomperts, R. and Mackenzie, M.J. (1979).
Contribution of urban runoff to hydrocarbon pollution. Water Pollution
Control, Vol. 51 (No. 8), pp. 2129-2138.
139. Jackson, D.A. (1991). Stopping rule in principal component analysis: A
comparison of heuristical and statistical approach. Ecology, Vol. 74 (No. 8),
pp. 2204-2214.
140. Jian-wei, Z., Bao-qing, S. and Cheng-qing, Y. (2007). Pollutant loads of
surface runoff in Wuhan city Zoo, an urban tourist area. Journal of
Environmental Sciences, Vol.19, pp. 464-468.
141. Karn, S.K. and Harada, H. (2001). Surface water pollution in three urban
territories of Nepal, India, and Bangladesh. Environmental Management, Vol.
28 (No. 4), pp. 483-496.
142. Kayhanian, M., Suverkropp, C., Ruby, A. and Tsay, K. (2007).
Characterization and prediction of highway runoff constituent event mean
concentration. Journal of Environment Management, Vol. 85, pp. 279–295.
143. Keller, H.R., Massart, D.L. and Brans, J.P. (1991). Multicriteria decision
making: a case study. Chemometrics and Intelligent laboratory systems, Vol.
11, pp. 175-189.
236
144. Kennedy, P. and Gadd, J. (2001). Preliminary examination of the nature of
urban roof runoff in New Zealand. Kingett Mitchell and Associates Ltd.
Prepared for Ministry of Transport.
145. Khalil, W.A., Goonetilleke, A., Kokot, S. and Carroll, S. (2004). Use of
chemometrics methods and multicriteria decision-making for site selection for
sustainable on-site sewage effluent disposal. Analytica Chimica Acta, Vol. 506,
pp. 41-56.
146. Kim, M., Chunga, H., Woo Y. and Kemper, M.S. (2007). A new non-invasive,
quantitative Raman technique for the determination of an active ingredient in
pharmaceutical liquids by direct measurement through a plastic bottle.
Analytica Chimica Acta, Vol. 587, pp. 200–207.
147. Kim, L.H., Zoh, K.D., Jeong, S., Kayhanian, M. and Stenstrom, M.K. (2006).
Estimating pollutant mass accumulation on highways during dry periods.
Journal of Environmental Engineering, Vol. 132 (No. 9), pp. 985-983.
148. Kokot, S. and Phuong, T.D. (1999). Element content of Vietnamese rice- Part
2, Multivariate data analysis. Analyst, Vol. 124, pp. 561-569.
149. Kokot, S., Grigg, M., Panayiotou, H. and Phuong, T.D. (1998). Data
interpretation by some common chemometrics methods. Electroanalysis, Vol.
10 (No. 16), pp. 1081-1088.
150. Kokot, S. and Yang, P. (1995). Comparison of thermogravimetric and
differential scanning calorimetric results for cellulosic fabrics by
chemometrics. Analytica Chemica Acta, Vol. 304, pp. 297-305.
151. Kronvang, B. (1992). The export of particulate matter, particulate phosphorus
and dissolved phosphorus from two agricultural river basins: implications on
estimating the non-point phosphorus load. Water Research, Vol. 26 (No.10),
pp. 1347-1358.
237
152. Lahdelma, R., Miettinen, K. and Salminen. (2003). Ordinal criteria in
stochastic multicriteria acceptability analysis (SMMA). European Journal of
Operational Research, Vol. 147, pp. 117-127.
153. Latimer, J.S., Hoffman, E.J., Hoffman, G., Fasching, J.L. and Quinn, J.G.
(2004). Sources of petroleum hydrocarbons in urban runoff. Water, Air and
Soil Pollution, Vol. 52 (No.1-2), pp. 1-21.
154. Latimer, J.S. (1984). Thesis-Department of Chemistry, University of Rhode
Island.
155. Laws, J.O. (1941). Measurements of the fall velocity of water drops and
raindrops. Transactions of the American Geophysical Union, Vol. 22, pp. 709-
721.
156. Lee, H., Swamikannu, X., Radulescu, D. and Kim, S. (2007). Design of
stormwater monitoring programs. Water Research, pp. 4186-4196.
157. Lee, J. G. and Heaney, J. P. (2003). Estimation of urban imperviousness and its
impacts on stormwater systems. Journal of Water Resources Planning and
Management, Vol 129 (No. 5), pp. 419-426.
158. Lee, H. and Stenstorm, K. (2005). Utility of stormwater monitoring. Water
Environmental Research, Vol. 77 (No. 3), pp. 219-228.
159. Lee, J. H. and Bang, K.W. (2000). Characterisation of urban stormwater runoff.
Water Research, Vol.34, pp. 1773-1780.
160. Lewis, J. (1996). Turbidity- controlled suspended sediment sampling for
runoff-event load estimation. Water Resources Research, Vol. 32 (No. 7), pp.
2299-2310.
238
161. Librando, V., Magazzu, G. and Puglisi, A. (1995). Multivariate micropollutants
analysis in marine waters. Water Science Technology, Vol. 32 (No. 9-10), pp.
341-348.
162. Lim, M.C.H., Ayoko, G.A., Morawska, L., Ristovski, Z.D., Jayaratne, E.R. and
Kokot, S. (2006). A comparative study of the elemental composition of the
exhaust emissions of cars powered by liquefied petroleum gas and unleaded
petrol. Atmospheric Environment, Vol. 40, pp. 3111-3122.
163. Line, D.E., White, N.M., Osmond, D.L., Jennings, G.D. and Mojonnier, C.B.
(2002). Pollutant export from various land uses in the upper neuse river basin,
Water Environment Research, Vol. 74, pp. 100-108.
164. Loch, R.J., Robotham, B.G., Zeller, L., Masterman, N., Orange, D.N., Bridge,
B.J., Sheridan, G. and Bourke, J. J. (2001). A multi-purpose rainfall simulator
for field infiltration and erosion studies. Australian Journal of Soil Research,
Vol.39, pp. 599-610.
165. Loch, R.J. (1982). Rainfall simulator methodology: concepts for realistic
research. Soil conservationist Queensland department of primary industries.
166. Macdonald, A.M., Edwards, A.C., Pugh, K.B. and Balls, P.W. (1995). Soluble
nitrogen and phosphorus in the river Ythan system, UK, Annual and seasonal
trends. Water Research, Vol.29 (No. 3), pp. 837-846.
167. Maestri, B. and Lord, B.N. (1987). Guide for mitigation of highway stormwater
runoff pollution: Science of the Total Environment, Highway Pollution. In:
Proceedings of the 2nd International Symposium, London, pp. 467-476.
168. Makepeace, D.K., Smith, D.W. and Stanley, S.J. (1995). Urban stormwater
quality, summary of contaminant data. Critical Reviews in Environmental
Science and Technology, Vol. 25 (No. 2), pp. 93-139.
239
169. Malley, D.F., Yesmin, L., Wray, D. and Edwards, S. (1999). Application of
near-infrared spectroscopy in analysis of soil mineral nutrients.
Communications in soil science and plant analysis, Vol. 30 (No. 7-8), pp. 999-
1012.
170. Mallin, M.A., Johnson, V.L. and Ensign, S. (2008). Comparative impacts of
stormwater runoff on water quality of an urban, a suburban and a rural stream.
Environmental Monitoring and Assessment.
171. Malvern Instrument Ltd. (1997). Sample dispersion and refractive index guide,
MAN 0079, U.K,
172. Mark, S.D. (1996). Stormwater discharges: Regulatory compliance and best
management practices, Published in CRC Press. ISBN 1566701988.
173. 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, Vol. 32 (No.
4), pp. 815-827.
174. Martens, H. (2001). Reliable and relevant modelling of real world data: a
personal account of the development of PLS regression. Chemometrics and
Intelligent Laboratory Systems, Vol. 58, pp. 85-95.
175. Martin, C., Ruperd, Y. and Legret, M. (2007). Urban stormwater drainage
management: The development of a multi criteria decision aid approach for
best management practices. European Journal of Operational Research, Vol.
181, pp. 338-349.
176. Martinez, A.A.M. (2005). Stormwater characteristics as described in the
national stormwater quality database. Thesis- University of Alabama,
Tuscaloosa, Alabama.
240
177. Martin, G.R., Smoot, J.L. and White, K.D. (1992). A comparison of surface-
grab and cross sectionally integrated stream- water-quality sampling methods.
Water Environment Research, Vol. 64 (No. 7), pp. 866-876.
178. Miguntanna, N. (2009). Nutrients build-up and wash-off processes in urban
land uses. Thesis (Unpublished)-Queensland University of Technology,
Brisbane Australia.
179. Milne, J. (2002). State of the Environment Surface Water Quality Monitoring:
A Review. Resource Science ISBN 1-877265-19-5.
180. Mosley, L.M. and Peake, B.M. (2001). Partitioning of metals (Fe, Pb, Cu, Zn)
in urban runoff from the Kaikorai Valley, Dunedin, New Zealand. New
Zealand Journal of Marine and Freshwater Research, Vol. 35, pp. 615-624.
181. Neary, V.S., Neel, C. and Dewey, J.B. (2002). Pollutant wash- off and loading
from parking lots in Cookeville, Tennessee. Urban drainage.
182. Nevado, J.J.B., Flores, R., Llerena, M.J.V. (1998). Simultaneous
spectrophotometric determination of Tartrazine, sunset yellow and ponceau 4R
in commercial products by PLS and principal component regression
multivariate calibration methods. Fresenius' Journal of Analytical Chemistry,
Vol. 361, pp. 465-472.
183. Nguyen, T.A., Kokot, S., Ongarato, D.M. and Wallace, G.G. (1999). The use
of chronoamperometry and chemometrics for optimization of conducting
polymer sensor arrays. Electroanalysis, Vol. 11 (No. 8), pp. 1327-1332.
184. Nielson, E. J. and Booth, D. B. (2002). Sediment sources in an urbanising,
Mixed land- use watershed. Journal of Hydro logy, Vol. 264, pp. 51-68.
185. Novotny, V. and Chesters, B. (1981). Handbook on nonpoint pollution:
Sources and management. Van Nostrand and Reinhold Company, New York.
241
186. Novotny, V. (1995). Nonpoint pollution and urban stormwater management,
Water quality management Library, Vol. 9, Techonomic Publishing Company,
Inc., Pennsylvania 17604, USA.
187. Novotny, V. (2003). Water quality: Diffuse pollution and watershed
management, Second Edition.
188. Novotny, V., Sung, H.M., Bannerman, R. and Baum, K. (1985). Estimating
nonpoint pollution from small urban watersheds. Journal Water Pollution
Control Federation, Vol. 57 (No. 4), pp. 339-348.
189. Ongley, E.D., Bynoe, M.C. and Percival, J.B. (1981). Highway stormwater
runoff quality: development of surrogate parameter relationships. Canadian
Journal of Earth Sciences, Vol. 18, pp. 1365-1379.
190. Packman, J.J., Comings, K.J. and Booth, D.B. (1999). Using turbidity to
determine total suspended solids in urbanizing streams in the Puget lowlands.
Managing Change in Water Resources and the Environment, Canadian Water
Resources Association annual meeting, Vancouver, BC, pp. 27-29.
191. Parker, N., Giglio, D., Gardner, T., Goonetilleke, A., Egodawatta, P., Thomas,
E., Gardiner, R. and Tan, C.K. (2008). Measuring the effectiveness of water
sensitive urban design (WSUD) treatment trains. Proceedings of Stormwater
Industry Association Conference, 8-11th July, Holiday Inn, Surfers Paradise,
Queensland, Australia.
192. Parks, S.J. and Baker, L.A. (1997). Sources and organic carbon in an Arizona
river- reservoir system. Water Research, Vol. 31 (No. 7), pp. 1751-1759.
193. Pazwash, H., Boswell, S.T., (1997). Management of runoff conservation and
reuse. Proceedings of the 24th Annual Water Resource Planning and
Management Conference, Houston, TX, ASCE, pp. 784-789.
242
194. Pechacek, L.D. (1994). Urban runoff based on landuse and particle size.
Conference Proceeding Paper, American Society of Civil Engineers, pp. 1242-
1246.
195. Pitt, R. (1979). Demonstration of nonpoint pollution abatement through
improved street cleaning practices. U.S. Environmental Protection Agency,
Report No. EPA/600/2-79-161.
196. Pitt, R., Field, R., Lalor, M. and Brown, M. (1995). Urban stormwater toxic
pollutants: Assessment, sources, and treatability. Water Environment Research,
Vol. 67 (No. 3), pp. 260-75.
197. Pitt, R., Bannerman, R., Clark, S. and Williamson, D. (2004). Sources of
pollutants in urban areas (Part 2)- Recent sheetflow monitoring. Effective
modeling of urban water systems, (Eds.) James, W., Irvine, K.N., McBean,
E.A. and Pitt, R.E., Monograph 13, pp. 485-506.
198. Polkowska, Z., Gorecki, T. and Namiesnik, J. (2002). Quality of roof runoff
waters from an urban region (Gdansk, Poland). Chemosphere, Vol. 49, pp.
1275-1283.
199. Pommer, L., Fick, J., Sundell, J., Nilsson, C., Sjostrom, M., Stenberg, B. and
Andersson, B. (2004). Class separation of buildings with high and low
prevalence of SBS by principal component analysis. Indoor air, Vol. 14, pp.
16-23.
200. Pouraghniaei, M.J. (2002). Effects of urbanisation on quality and quantity of
water in the watershed. Msc of Watershed Management, Natural Resources
Research Center of Semnan, Semnan Province, Iran.
201. Power, J.F. and Schepers, J.S. (1989). Nitrate contamination of groundwater in
North America. Agriculture, Ecosystems and Environment, Vol. 26 (No. 3-4),
pp. 165-187.
243
202. Pringle, T. (1998). Selecting roof cladding. A guide to choosing roofing
materials, BRANZ, Welligton.
203. PRS-Sirius Version 7.1 Software - User Guide.
204. Purcell, D.E., Leonard, G.J., O’Shea, M. and Kokot, S. (2005). A
chemometrics investigation of sugarcane plant properties based on the
molecular composition of epicuticular wax. Chemometrics and Intelligent
Laboratory Systems, Vol. 76, pp. 135-147.
205. Quek, U. and Forster, J. (1993). Trace metals in roof runoff. Water, Air and
Soil Pollution, Vol. 68, pp. 373-389.
206. Rahmat, S.N.B. (2005). Modelling of nonpoint source pollution from
residential and commercial catchments in Skudai, Johor. Thesis- University of
Teknologi Malaysia.
207. Rahman, A., Thomos, E., Bhuiyan, S. and Goonetilleke, A. (2002). Modelling
pollutant wash-off from South East Queensalnd catchments Australia.
Proceedings of International Water Association 3rd World water congress.
208. Randall, D. (2003). An introduction to partial least squares regression. SAS
institute, Cary, NC.
209. Rickson, R.J. (2001). Experimental techniques for erosion studies: rainfall
simulation.
210. Roberts, A. R. and Withers, P. (2004). Statisti XL, Verison 1.4.
211. Robian, A., Striebel, T. and Herrmann, R. (1997). Modelling of dissolved and
particle-bound pollutants in urban street runoff. Water Science and
Technology, Vol. 36 (No. 8-9), pp. 77-82.
244
212. Rogge, W.F., Hlldemann, L.M., Mazurek, M.A. and Cass, G.R. (1993).
Sources of fine organic aerosol-road dust, Tire debris and Organometallic
brake lining dust: Roads as sources and sinks. Environmental Science and
Technology, Vol. 27 (No. 9), pp. 1892-1904.
213. Roger, S., Montrejaud, V.M., Andral, M.C., Herrmans, L. and Fortune, J.P.
(1998). Mineral, physical and chemical analysis of the solid matter carried by
motorway runoff water. Water Resources Research, Vol. 34 (No. 4), pp. 1119-
1125.
214. Rosewell, C.J. (1986). Rainfall kinetic energy in eastern Australia. Journal of
Climate and Applied Meteorology, Vol.25, pp. 1695-1701.
215. Sansalone, J.J., Buchberger, S.G. and Al-Abed, S.R. (1996). Fractionation of
heavy metals in pavement runoff. The Science of the Total Environment,
189/190, pp. 371-378.
216. Sansalone, J.J., Koran, J.M., Smithson, J.A. and Buchnerger, S.G. (1998).
Physical characteristics of urban roadway solids transported during rain events.
Journal of Environmental Engineering, pp. 427-440.
217. Sara, D.L., Wong, T.H.F. and Chesterfield, C.J. (2002). Water sensitive urban
design- A stormwater management perspective. Cooperative research centre for
catchment hydrology, Melbourne water Corporation, Industry report, Report
02/10.
218. Sartor, J.D. and Boyd, B.G. (1972). Water pollution aspects of street surface
contaminants. Report No. EPA- R2-72/081, US Environmental Protection
Agency, Washington, DC, USA.
219. Sartor, J.D., Boyd, B.G. and Agardy, F.J. (1974). Water pollution aspects of
street surface contaminants. U.S. Environmental Protection Agency,
Washington, DC, USA, Report No. EPA- R2-72/081.
245
220. Saulius, T., Willeke, K., Reponen, T., Grinshpuna, S.A. and Friedman, W.
(2001). Comparison of filter bag, cyclonic and wet dust collection methods in
vacuum cleaners. American Industrial Hygiene Association, Vol. 62, pp. 573-
583.
221. Scholes, L., Revitt, D.M. and Ellis, J.B. (2007). A systematic approach for the
comparative assessment of stormwater pollutant removal potentials. Journal of
Environmental Management (Article in Press).
222. Schuler, T. (1997). Impact of suspended and deposited sediment. Technical
note 86, pp. 443-444.
223. Schueler, T.R., P.A. Cumble, and M.A. Heraty (1991). Current assessment of
urban best management practices: Techniques for reducing non-point source
pollution in the coastal zones. (Review Draft). Metropolitan Washington
Council of Governments. Washington, D.C.
224. Seaburn, G.E. (1969). Effect of urban development on direct runoff to East
Meadow Brook, Nassau Country, Long Island, New York. US Geological
Survey Professional Paper 627-B, pp. 1-14.
225. Settle, S., Goonetilleke, A. and Ayoko, G.A. (2007). Determination of
surrogate indicators for phosphorus and solids in urban stormwater:
Application of multivariate data analysis techniques. Water, Air and Soil
Pollution, Vol. 182 (No. 1-4), pp. 149-161.
226. Settle, S. and Goonetilleke, A. (2001). Surrogate indicator assessment in
stormwater for solids and phosphorus. Physical infrastructure Centre, Research
Monograph 01-01, Queensland university of Technology, Brisbane, Australia.
227. Shaheen, D.G. (1975). Contributions of urban roadways to water pollution. US
Environmental protection Agency, EPA-600/2-75-004.
246
228. Shigaki, F., Sharpley, A. and Prochnow, L. (2007). Rainfall intensity and
phosphorus source effects on phosphorus transport in surface runoff from soil
trays. Science of the Total Environment, Vol. 373, pp. 334-343
229. Shuster, W.D., Bonta, J., Thurston, H., Warnemuende, E. and Smith, D.R.
(2005). Impacts of impervious surface on watershed hydrology: A review.
Urban Water Journal, Vol. 2 (No. 4), pp. 263-275.
230. Sonzogni, W.C., Chesters, G., Coote, D.R., Jeffs, D.N., Konrad, J.C., Ostry,
R.C. and Robinson, J.B. (1980). Pollution from land runoff. Environmental
Science and Technology, Vol. 14 (No. 2), pp. 148-153.
231. Steuer, Jeffrey, William, S., Hornewer, N. and Prey, J. (1997). Sources of
contamination in an urban basin in Marquette, Michigan and an analysis of
concentrations, loads, and data quality, U.S. Geological Survey Water-
Resources Investigations Report 97-4242, Middleton, Wisconsin.
232. Sun, H. (2004). A universal molecular descriptor system for prediction of
LogP, LogS, LogBB and absorption. Journal of Chemical Information and
Modelling, Vol. 44 (No. 2), pp. 748-757.
233. Sutherland, R.A., Tack, F.M.G., Tolosa, C.A. and Verloo, M.G. (1998).
Operationally defined metal fractions in road deposited sediment, Honolulu,
Hawaii. Journal of Environmental Quality, Vol. 29 (No. 5), pp. 1431-1439.
234. Tai, Y.L. (1991). Physical and chemical characterisation of street dust and
dirt from urban areas. Thesis- Pennsylvania State University.
235. Taylor, R. (1998). Stormwater management for service stations in public
drinking water source areas.
236. Taylor, G.D., Fletcher, T.D., Wong, T.H.F., Breen, P.F. and Duncan, H.P.
(2005). Nitrogen composition in urban runoff- implications for stormwater
management. Water Research, Vol. 39, pp. 1982-1989.
247
237. Tiefenthaler, L.L. and Kenneth, C.S. (2001). Effects of rainfall intensity and
duration on first flush of stormwater pollutants. SCCWRP Annual report 2001-
2002.
238. Thomas, P.R. and Greene, G.R. (1993). Rainwater quality from different roof
catchments. Water Science and Technology, Vol. 28, pp. 291-299.
239. Thomson, N.R., Mcbean, E.A., Snodgrass, W. and Monstrenko, I.B. (1997).
Highway stormwater Quality: Development of surrogate parameter
relationships. Water, Air, and Soil Pollution, Vol. 94, pp. 307-347.
240. Thornton, R.C. and Saul, A.J. (1986). Some quality characteristics of combined
sewer flows. Public Health Engineering, Vol. 14 (No. 3), pp. 35-39.
241. Tsihrintzis, V.A. and Hamid, R. (1997). Modelling and management of urban
stormwater runoff quality: A Review. Water Resources Management, Vol. 11,
pp. 137-164.
242. Tyler, B.J., Rayal, G. and Castner, D.G. (2007). Multivariate analysis strategies
for processing TOF- SIMS images of biomaterials. Biomaterials, Vol. 28, pp.
2412-2423.
243. Ujevic, I., Odzak, N. and Baric, A. (2000). Trace metal accumulation in
different grain size fractions of the sediments from a semi- enclosed bay
heavily contaminated by urban and industrial wastewaters. Water Research,
Vol. 34 (No. 11), pp. 3055-3061.
244. US Department of Transportation, Federal Highway Administration. Guidance
Manual for Monitoring Highway runoff Water Quality. Publication No.
FHWA- EP-01-022, pp.1-5.
245. US EPA, 1983, Method 365.4: Phosphorous, total (Colorimetric, automated,
block digester AA II). US EPA Environmental Protection Agency,
Washington, DC. EPA/600/4-79/020.
248
246. US EPA, 1993, Method 351.2: Determination of total kjeldahl nitrogen by semi
automated colorimetry. US EPA Environmental Protection Agency,
Washington, DC. EPA/600/4-79/020.
247. US. Environmental Protection Agency, (1993). Natural wetlands and urban
stormwater potential impacts and management.
248. US Environmental Protection Agency (1999). Protocol for developing nutrient
TMDLs. U.S. Environmental Protection Agency, Report No. EPA /841/B-
99/007.
249. US Environmental Protection Agency (1997). Monitoring guidance for
determining the effectiveness of non-point source controls. U.S. Environmental
Protection Agency, Report No. EPA /841/B-96/004.
250. US Federal Highway Administration (2001). Guidance manual for monitoring
highway runoff water quality. U.S. department of transformation, Federal
highway administration, publication No. FHWA-EP-01-022.
251. US Environmental Protection Agency (2002). Urban stormwater BMP
performance monitoring. A guidance for meeting the national stormwater BMP
data base requirements.
252. Urbonas, B. (1994). Parameters to report with BMP monitoring data. In
stormwater NPDES monitoring needs. American Society of Civil Engineers.
Proceedings of the Engineering Foundation Conference on Storm Water
Monitoring. August 7-12, Crested Butte, CO.
253. Urbonas, B. and P. Stahre (1993). Stormwater- Best management practices and
detention. Prentice Hall. Englewood Cliffs, NJ.
254. Urbonas, B. (2000). Assessment of Stormwater Best Management Practice
Effectiveness. (Chapter7). Innovative urban wet weather flow management
systems, Field R, Heaney, J.P., Pitt, R. (editors). Economic publishing Co, Inc.,
Lancaster, P.A. pp. 255-300.
249
255. Uunk and Ven (1987). Water quantity and quality in the new town of Lelystad
(M). In: Effects of land use on fresh water: agriculture, forestry, mineral
exploitation, urbanisation (Solbe, J. F., De L. G., ed.), WRc/Ellis Horwood,
Chichester.
256. Van Metre, P.C., Mahler, B.J. and Furlong, E.T. (2000). Urban sprawl leaves
its PAH signature. Environmental Science and Technology, Vol. 34 (No. 19),
pp. 4064-4070.
257. Van Metre, P.C. and Mahler, B.J. (2003). The contribution of particles washed
from rooftops to contaminant loading to urban streams. Chemosphere, Vol. 52,
pp. 1727-1741.
258. Vaze, J., Chiew, F.H.S., Suryadi, L. and Khanal, K. (2000). Pollutant
accumulation on an urban road surface. Urban Water, Vol. 4, pp. 265-270.
259. Vaze, J. and Chiew, F.H.S. (2002). Experimental study on pollutant
accumulation on an urban road surface. Urban Water, Vol. 4, pp. 379-389.
260. Vaze, J. and Chiew, F.H.S. (2004). Nutrient Loads associated with different
sediment sizes in urban stormwater and surface pollutants. Journal of
environmental Engineering, Vol. 130 (No. 4), pp. 391-396.
261. Vega, M., Pardo R., Barrado E. and Deban, L. (1998). Assessment of seasonal
and polluting effects on the quality of river water by exploratory data analysis.
Water Research, Vol. 32 (No. 12), pp. 3581-3592.
262. Viklander, M. (1998). Particle size distribution and metal content in street
sediments. Journal of Environmental Engineering, Vol. 124 (No. 8), pp. 761-
766.
263. Waananen, A.O. (1969). Urban effects of water yield. Proceedings of the
second water resources symposium, Effects of watershed changes on stream
flow, Austin.
250
264. Walker, T.A. and Wong, T.H.F. (1999). Effectiveness of street sweeping for
stormwater pollution control, Technical report, Report 99/8, Cooperative
research centre for catchment hydrology.
265. Walling, D.E. and Kane, P. (1982). Temporal variation of suspended sediment
properties. Recent developments in the explanation and prediction of erosion
and sediment yield, proceedings of the Exeter Symposium, IAHS publ.no.137.
266. Walsh, C.J. (2000). Urban impacts on the ecology of receiving waters: a
framework for assessment, conservation and restoration. Hydrobiologia, Vol.
431, pp. 107-114.
267. Wang, T., Spyridakis, D.E., Mar, B.W. and Horner, R.R. (1981). Transport
deposition and control of heavy metals in highway runoff. Report No.10,
Washington state Dept. of Transport, Department of Civil Engineering.
268. Wang, X-C., Zhang, Y-X and Chen, R.F. (2001). Distribution and partitioning
of Polycyclic Aromatic Hydrocarbons (PAHs) in different size fractions in
sediments from Boston Harbor, US. Marine Pollution Bulletin, Vol. 42 (No.
11), pp. 1139-1149.
269. Warren, N., Allan, I.J., Carter, J.E., House, W.A. and Parker, A. (2003).
Pesticides and other micro-organic contaminants in freshwater sedimentary
environments- A review. Applied Geochemistry, Vol. 18, pp. 159-194.
270. Weeks, C.R. (1981). Pollution in urban stormwater runoff. Second
International Conference on Urban Storm Drainage, pp. 39-47.
271. Westco Scientific Instruments, SmartChem Discreate Analyser operator
manual.
272. Williams, P.C. (1987). Variables affecting near-infrared reflectance
spectroscopic analysis. Near-infrared technology in the agricultural and food
251
industries, Chapter 7, Grain research laboratory, Canadian grain commission,
Winnipeg, Manitoba, Canada, pp. 143-167.
273. Williamson, R.B. (1986). Urban stormwater quality: Comparison of three New
Zealand catchments. New Zealand Journal of Marine and Freshwater Research,
Vol. 20, pp. 315-328.
274. Wold, S., Sjostrom, M. and Eriksson, L. (2001). PLS-regression: a basic tool of
chemometrics. Chemometrics and intelligent laboratory systems, Vol. 58, pp.
109-130.
275. Wong, T.H.F. (2001). A changing paradigm in Australian urban stormwater
management. A paper submitted for the 2nd South Pacific Stormwater
conference, New Zealand.
276. Yaziz, M.I., Gunting, H., Sapari, N. and Ghazali, A.W. (1989). Variations in
rainwater quality from roof catchments. Water resources, Vol. 23 (No. 6), pp.
761-765.
277. Yufen, R., Xiaoke, W., Zhiyun, O., Hua, Z., Xiaonan, D. and Hong, M. (2008).
Stormwater runoff quality from different surfaces in an urban catchment in
Beijing, China. Water Environment Research, Vol. 80 (No. 8), pp. 719-724.
278. Zafra, C.A., Temprano, J. and Tejero, I. (2008). Particle size distribution of
accumulated sediments on an urban road in rainy weather. Environmental
Technology, Vol. 29, pp. 571-582.
279. Zeng, X. and Rasmussen, T.C. (2005). Mulitvariate statistical characterisation
of water quality in Lake Lanier, Georgia, USA. Journal of Environmental
Quality, Vol. 34. pp. 1980-1991.
280. Zhang, G., Ni, Y., Churchill, J. and Kokot, S. (2006). Authentication of
vegetable oils on the basis of their physico-chemical properties with the aid of
chemometrics. Talanta Vol. 70, pp. 293-300.
252
281. Zoppou, C. (2001). Review of urban stormwater models. Environmental
Modelling and Software, Vol. 16 (No. 3), pp. 195-231.
253
Appendix A
Rainfall Simulator Calibration Data
254
255
Rainfall simulator calibration for rainfall intensity
Figure 1 shows the container positions used for the rainfall simulator calibration. The
rainfall was simulated for a number of known settings of the control box for a
duration of 5 minutes. The amount of water collected in all the containers was
measured and the rainfall intensity in terms of depth was calculated. Table 1 shows
the selected control box settings to generate the selected rainfall intensities in the
research undertaken.
Figure 1- The position of the containers
Cup position
256
Table 1- Calculation of rainfall intensity from the measured water volume
Control box setting Cup number
Volume mL
Intensity mm/hr
Duration min
1 10 19.5 5 2 12 23.4 5 3 14 27.3 5 4 10 19.5 5 5 8 15.6 5 6 15 29.2 5 7 16 31.2 5 8 20 38.9 5 9 16 31.2 5 10 12 23.4 5 11 12 23.4 5 12 16 31.2 5 13 14 27.3 5 14 14 27.3 5 15 14 27.3 5 16 4 7.8 5 17 6 11.7 5 18 10 19.5 5 19 10 19.5 5
A1 (20 mm/hr)
20 8 15.6 5 1 14 27.3 5 2 14 27.3 5 3 18 35.1 5 4 12 23.4 5 5 10 19.5 5 6 28 54.5 5 7 26 50.6 5 8 28 54.5 5 9 26 50.6 5 10 20 38.9 5 11 22 42.8 5 12 24 46.7 5 13 30 58.4 5 14 24 46.7 5 15 30 58.4 5 16 10 19.5 5 17 10 19.5 5 18 25 48.7 5 19 24 46.7 5
H1 (40 mm/hr)
20 20 38.9 5
257
Table 1 Contd.
Control box setting Cup
number Volume
mL Intensity mm/hr
Duration min
1 24 46.7 5 2 24 46.7 5 3 32 62.3 5 4 22 42.8 5 5 16 31.2 5 6 92 179.1 5 7 38 74.0 5 8 46 89.6 5 9 38 74.0 5 10 36 70.1 5 11 44 85.7 5 12 36 70.1 5 13 42 81.8 5 14 36 70.1 5 15 36 70.1 5 16 12 23.4 5 17 14 27.3 5 18 26 50.6 5 19 20 38.9 5
J2 (65 mm/hr)
20 22 42.8 5 1 30 58.4 5 2 36 70.1 5 3 42 81.8 5 4 28 54.5 5 5 22 42.8 5 6 128 249.2 5 7 50 97.4 5 8 56 109.0 5 9 50 97.4 5 10 44 85.7 5 11 52 101.3 5 12 46 89.6 5 13 52 101.3 5 14 46 89.6 5 15 50 97.4 5 16 16 31.2 5 17 20 38.9 5 18 36 70.1 5 19 28 54.5 5
K3 (86 mm/hr)
20 26 50.6 5
258
Table 1 Contd.
Control box setting Cup
number Volume
mL Intensity mm/hr
Duration min
1 44 85.7 5 2 50 97.4 5 3 56 109.0 5 4 38 74.0 5 5 26 50.6 5 6 120 233.7 5 7 76 148.0 5 8 78 151.9 5 9 72 140.2 5 10 60 116.8 5 11 80 155.8 5 12 70 136.3 5 13 80 155.8 5 14 70 136.3 5 15 66 128.5 5 16 26 50.6 5 17 32 62.3 5 18 50 97.4 5 19 44 85.7 5
L6 (115 mm/hr)
20 40 77.9 5 1 54 105.2 5 2 64 124.6 5 3 74 144.1 5 4 46 89.6 5 5 30 58.4 5 6 90 175.3 5 7 96 186.9 5 8 84 163.6 5 9 90 175.3 5 10 74 144.1 5 11 90 175.3 5 12 90 175.3 5 13 100 194.7 5 14 88 171.4 5 15 86 167.5 5 16 28 54.5 5 17 40 77.9 5 18 64 124.6 5 19 52 101.3 5
M4 (135 mm/hr)
20 52 101.3 5
Table 2- Calculation of median drop size using flour pellet method
Sieve size
(mm) Weight of
pellets (g)
Weight of a single pellet (mg)
Number of pellets
Calibration ratio
Mass of a
water drop (mg)
Volume (cm3)
Average drop
diameter (mm)
Total volume (mm3)
% of total
volume
>4.75 1.338 57.50 23 1.25 71.95 0.072 5.16 1.67 8.61 4.75-3.35 3.183 21.85 146 1.28 27.90 0.028 3.76 4.07 20.91 3.35-2.36 4.361 13.57 321 1.27 17.29 0.017 3.21 5.56 28.59 2.36-1.68 3.799 4.60 826 1.21 5.58 0.006 2.20 4.61 23.71 1.68-1.18 2.007 1.87 1070 1.12 2.09 0.002 1.59 2.24 11.52 1.18-0.85 1.030 0.49 2098 0.79 0.39 0.000 0.90 0.81 4.17
<0.85 0.730 0.22 3266 0.66 0.15 0.000 0.66 0.48 2.49
Table 2- Contd:
According to Table 2; D50- Median drop size diameter =2.45mm
Average drop
diameter (mm)
Percentage of total volume
(%)
Cumulative volume (%)
0.66 2.49 2.49 0.90 4.17 6.66 1.59 11.52 18.18 2.20 23.71 41.89 3.21 28.59 70.48 3.76 20.91 91.39 5.16 8.61 100.00
261
Table 3- Calculation of kinetic energy
Average drop
diameter (mm)
Mass of a drop (mg)
Terminal velocity
(m/s)
Kinetic energy per
drop (J)
Number of drops in
each class
Kinetic energy of all drops
(J)
5.16 71.95 8.71 0.00273 23 0.064 3.76 27.90 8.70 0.00106 146 0.154 3.21 17.29 8.26 0.00059 321 0.189 2.20 5.58 6.98 0.00014 826 0.112 1.59 2.09 5.97 0.00004 1070 0.040 0.90 0.39 4.23 0.00001 2098 0.007 0.66 0.15 3.71 0.00001 3266 0.003
Total kinetic energy of the rain drops = 0.57 J
Sample area = 34.5*45.5 cm2
= 0.16 m2
Duration of the sample was taken = 3.2s
Rainfall intensity of the sample was taken = 159.3mm/hr
According to Kinnell (1987)
Kinetic energy per unit depth of rain = (0.57x3600)/ [159.3x3.2x 0.16]
= 25.63 J/m2/mm
262
263
Appendix B
Raw Data from Field Trials
264
265
Original build-up test results for road surfaces Table 1- Total build-up samples
Road Site Sample
date Volume (L) Antecedent
dry days Drumbeat street 3/07/2008 7.88 14
Ceil Circuit 31/07/2008 6.64 7
Particle size distribution (%) Road Site <1 µm 1-75 µm 75-150 µm 150-300 µm >300 µm
Drumbeat street 3.35 70.87 18.17 4.82 2.79 Ceil Circuit 1.16 48.42 27.33 13.62 9.47
Road Site TSS
(mg/L) TOC
(mg/L) NO2
-
(mg/L) NO3
- (mg/L)
TKN (mg/L)
TN (mg/L)
TPO43-
(mg/L) TP
(mg/L) Drumbeat street 688 27.760 0.012 0.722 8.077 8.811 2.699 2.968
Ceil Circuit 266.4 8.737 <0.001 0.370 3.856 4.226 0.714 0.822 Note: TSS- Total suspended solids; TOC- Total organic carbon; NO2
-- nitrite-nitrogen; NO3-- nitrate- nitrogen; TKN- Total kjeldahl nitrogen; TN- Total nitrogen;
TPO43-- Total Phosphates; TP- Total phosphorus.
266
Table 2- Wet sieved build-up samples; DR-Drumbeat Street; CE-Ceil Circuit
Sample name TSS
(mg/L) TOC
(mg/L) NO2-
(mg/L) NO3-
(mg/L) TKN
(mg/L) TN
(mg/L) TPO4
3- (mg/L)
TP (mg/L)
DR-BU>300 141.600 3.514 <0.001 0.109 0.600 0.709 0.523 0.530
DR-BU-300-150 154.400 3.883 <0.001 0.120 0.086 0.205 0.954 0.984
DR-BU-150-75 275.200 8.160 <0.001 0.095 1.873 1.969 1.422 1.443
DR-BU-75-1 588.800 17.470 <0.001 0.022 2.007 2.027 1.559 1.568
DR-BU<1 256.000 12.100 0.001 0.696 1.053 1.750 0.035 0.096
CE-BU>300 73.600 2.198 <0.001 0.029 0.085 0.114 0.394 0.980
CE-BU-300-150 43.200 4.453 <0.001 0.032 0.023 0.055 0.603 0.861
CE-BU-150-75 400.400 7.474 <0.001 0.058 1.767 1.826 0.905 1.054
CE-BU-75-1 11.200 4.130 <0.001 0.109 0.266 0.375 1.082 1.152
CE-BU<1 260.000 3.795 <0.001 0.254 0.099 0.352 0.123 0.130 Notes: DR- Drumbeat Street; CE- Ceil Circuit; BU- Build-up. TSS- Total suspended solids; TOC- Total organic carbon; NO2
-- nitrite-nitrogen; NO3-- nitrate-
nitrogen; TKN- Total kjeldahl nitrogen; TN- Total nitrogen; TPO43-- Total Phosphates; TP- Total phosphorus.
267
Table 3- Original build-up test results for roof surfaces; T-Tile roof surface; S-Steel Roof surface
Sampling Episode Sample name Sample date Volume (L)
Antecedent dry days
T-1 18/09/2008 6.36 8 1 S-1 18/09/2008 6.44 8
T-2 24/09/2008 6.32 6 2 S-2 24/09/2008 6.46 6
T-3 30/09/2008 6.62 6 3 S-3 30/09/2008 7.52 6
Particle size distribution (%) Sampling Episode
Sample name
<1 µm 1-75 µm 75-150 µm 150-300 µm >300 µm T-1 2.50 57.9 19.22 8.48 11.91
1 S-1 0.00 73.28 2.33 4.54 19.85 T-2 2.48 54.69 18.67 7.03 17.13
2 S-2 1.03 59.08 3.41 1.16 35.33 T-3 3.45 61.85 19.49 9.3 5.91
3 S-3 3.99 59.33 13.64 7.89 15.15
268
Table 3 Cond:
Sampling Episode
Sample name
TSS (mg/L)
TOC
(mg/L)
NO2
- (mg/L)
NO3-
(mg/L) TKN
(mg/L) TN
(mg/L) TPO4
3-
(mg/L) TP
(mg/L) T-1 66.800 1.497 0.038 0.202 0.598 0.838 2.182 2.586
1 S-1 33.600 0.494 0.039 0.202 0.677 0.918 3.238 3.270 T-2 52.800 3.951 0.033 0.075 0.485 0.593 2.535 2.864
2 S-2 29.200 3.913 0.036 0.103 0.886 1.025 4.253 4.457 T-3 54.800 2.709 0.028 0.080 0.300 0.408 0.700 1.229
3 S-3 19.600 1.724 0.029 0.094 0.361 0.483 0.124 0.170 Note: T-Tile roof surface; S-Steel Roof surface; TSS- Total suspended solids; TOC- Total organic carbon; NO2
-- nitrite-nitrogen; NO3-- nitrate- nitrogen; TKN-
Total kjeldahl nitrogen; TN- Total nitrogen; TPO43-- Total Phosphates; TP- Total phosphorus.
269
Table 4- Wet sieved build-up samples Sampling Episode Sample name
TSS (mg/L)
TOC (mg/L)
NO2-
(mg/L) NO3
- (mg/L)
TKN (mg/L)
TN (mg/L)
TPO43-
(mg/L) TP
(mg/L) T-1-BU>300 6.800 0.773 0.038 0.123 0.151 0.312 0.245 0.347 S-1-BU>300 0.400 0.567 0.038 0.109 0.189 0.336 <0.03 <0.007 T-1-BU-300-150 1.600 0.928 0.034 0.127 0.203 0.364 0.275 0.332 S-1-BU-300-150 4.400 0.536 0.034 0.139 0.103 0.276 <0.03 <0.007 T-1-BU-150-75 26.000 5.948 0.030 0.129 1.513 1.672 0.075 0.090 S-1-BU-150-75 6.000 8.341 0.035 0.115 0.170 0.320 <0.03 <0.007 (T-1)-BU-75-1 11.600 2.680 0.005 0.046 0.250 0.301 0.024 0.022 (S-1)-BU-75-1 0.000 0.000 0.006 0.005 0.325 0.337 0.040 0.043 T-1-BU<1 46.000 4.671 0.032 0.156 0.765 0.953 0.253 0.354
1 S-1-BU<1 50.000 6.639 0.024 0.205 1.254 1.483 0.219 0.243 T-2-BU>300 6.000 0.741 0.011 0.065 0.530 0.606 0.015 0.018 S-2-BU>300 6.400 0.619 0.027 0.092 0.054 0.173 0.125 0.178 T-2-BU-300-150 8.800 0.750 0.026 0.090 0.258 0.373 0.036 0.040 S-2-BU-300-150 6.400 0.528 0.023 0.088 0.043 0.154 0.144 0.153 T-2-BU-150-75 30.800 1.149 0.028 0.092 0.257 0.377 <0.03 <0.03 S-2-BU-150-75 4.000 0.581 0.027 0.096 0.070 0.192 0.131 0.152 (T-2)-BU-75-1 0.000 0.938 0.010 0.000 0.261 0.271 0.212 0.076 (S-2)-BU-75-1 0.000 0.552 0.006 0.014 0.029 0.049 0.098 0.080 T-2-BU<1 38.000 3.161 0.026 0.105 1.352 1.483 0.372 0.532
2 S-2-BU<1 36.000 2.877 0.023 0.082 0.336 0.441 0.799 0.852 T-3-BU>300 52.000 0.582 0.030 0.085 0.128 0.243 0.173 0.183 S-3-BU>300 12.000 0.553 0.028 0.090 0.051 0.169 0.118 0.126 T-3-BU-300-150 50.000 0.626 0.023 0.088 0.093 0.204 0.102 0.122 S-3-BU-300-150 34.000 0.439 0.030 0.096 0.168 0.295 0.121 0.131 T-3-BU-150-75 124.000 0.674 0.032 0.097 0.089 0.218 0.125 0.146 S-3-BU-150-75 44.000 0.806 0.029 0.097 0.138 0.264 0.132 0.142 (T-3)-BU-75-1 156.000 0.360 0.005 0.012 0.215 0.232 0.022 0.030 (S-3)-BU-75-1 42.000 0.285 0.007 0.034 0.013 0.054 0.045 0.049 T-3-BU<1 26.000 1.828 0.023 0.084 0.370 0.478 0.141 0.143
3 S-3-BU<1 26.000 1.665 0.022 0.061 0.246 0.328 0.085 0.091
270
Original wash-off test results for road surfaces: T-Total samples; D- Filtrates Table 5- Wash-off test results- Drumbeat Street site
Sample name Identification Particle size distribution (%) Intensity(mm/hr)
and duration(min) <1 µm 1-75 µm
75-150 µm
150-300 µm
>300 µm
DR-20-0-5 DR20-1 0.51 40.81 6.60 1.64 50.44
DR-20-5-10 DR20-2 0.18 17.10 1.47 0.12 81.13
DR-20-10-15 DR20-3 0.10 7.46 0.59 0.02 91.83
DR-20-15-20 DR20-4 0.00 7.49 0.63 0.04 91.84
DR-20-20-25 DR20-5 0.21 10.79 6.51 4.52 77.97
DR-20-25-30 DR20-6 0.36 12.55 5.28 3.62 78.19
DR-20-30-35 DR20-7 0.00 9.25 2.85 3.19 84.72
DR-20-35-40 DR20-8 0.00 3.56 0.23 0.01 96.20
DR-40-0-5 DR40-1 1.31 61.55 9.26 6.19 21.69
DR-40-5-10 DR40-2 0.87 37.99 8.28 5.28 47.58
DR-40-10-15 DR40-3 0.62 18.72 2.91 0.74 77.01
DR-40-15-20 DR40-4 0.00 15.47 1.70 0.10 82.74
DR-40-20-25 DR40-5 0.74 20.08 3.35 1.11 74.72
DR-40-25-30 DR40-6 0.09 82.60 0.94 6.37 9.99
DR-40-30-35 DR40-7 1.61 79.18 1.25 9.37 8.59
DR-65-0-5 DR65-1 1.34 42.91 9.51 4.52 41.72
DR-65-5-10 DR65-2 0.83 31.19 14.29 4.89 48.81
DR-65-10-15 DR65-3 0.00 6.84 0.77 0.01 92.39
DR-65-15-20 DR65-4 1.36 46.36 12.52 1.81 37.86
DR-65-20-25 DR65-5 0.51 17.47 6.50 1.34 74.19
DR-65-25-30 DR65-6 0.00 13.18 1.39 0.06 85.37
DR-86-0-5 DR86-1 2.42 63.58 12.20 7.27 14.54
DR-86-5-10 DR86-2 1.02 30.04 5.60 1.04 62.30
DR-86-10-15 DR86-3 1.39 42.66 9.59 2.29 44.08
DR-86-15-20 DR86-4 0.99 32.93 7.76 1.00 57.33
DR-86-20-25 DR86-5 0.75 27.98 7.63 1.30 62.35
DR-115-0-5 DR115-1 1.27 41.56 10.14 11.87 35.16
DR-115-5-10 DR115-2 0.70 29.21 12.82 12.68 44.59
DR-115-10-15 DR115-3 0.47 18.57 9.54 3.00 68.42
DR-115-15-20 DR115-4 0.00 0.00 0.00 0.08 99.92
DR-135-0-5 DR135-1 1.95 60.46 17.04 3.75 16.81
DR-135-5-10 DR135-2 2.79 51.41 17.57 4.20 24.03
DR-135-10-15 DR135-3 2.45 34.96 21.83 10.66 30.10
DR-135-15-20 DR135-4 0.74 22.76 24.46 5.12 47.01
271
Table 5 Contd:
Identification pH EC
(µS/cm) TSS
(mg/L) TDS
(mg/L) TTU
(NTU) TOC
(mg/L) DOC
(mg/L) DR20-1 7.40 170.5 64.8 528.0 55.00 165.800 146.800
DR20-2 6.56 107.0 46.4 436.0 37.00 93.720 81.760
DR20-3 6.85 59.6 25.2 244.0 17.70 44.990 40.640
DR20-4 7.05 52.1 21.6 188.0 14.40 34.280 29.960
DR20-5 7.09 52.3 16.2 124.0 12.30 26.020 25.700
DR20-6 7.05 50.9 14.8 68.0 11.40 21.590 18.750
DR20-7 7.16 49.6 13.6 32.0 9.30 13.890 13.550
DR20-8 7.18 50.3 12.0 18.0 6.70 15.710 15.480
DR40-1 8.10 84.5 55.6 252.0 49.00 63.230 48.820
DR40-2 8.64 47.9 38.4 222.0 30.00 31.320 25.600
DR40-3 7.94 42.6 32.0 220.0 22.00 24.170 19.380
DR40-4 8.11 42.2 25.6 200.0 19.00 17.860 15.840
DR40-5 8.46 48.6 17.2 200.0 13.00 13.290 12.230
DR40-6 8.77 45.3 12.4 188.0 10.50 12.100 11.420
DR40-7 8.70 44.6 10.2 88.0 8.80 11.140 10.040
DR65-1 6.06 114.0 131.6 128.0 19.80 31.740 26.520
DR65-2 7.37 88.1 72.0 112.0 12.60 17.210 14.560
DR65-3 7.78 121.3 50.4 90.0 11.80 12.100 11.470
DR65-4 7.63 41.0 45.2 84.0 11.50 9.813 9.277
DR65-5 7.63 53.9 35.2 74.0 10.40 9.248 8.487
DR65-6 7.58 65.0 31.2 48.0 8.60 7.650 8.176
DR86-1 6.61 48.3 222.0 332.0 94.00 29.690 22.840
DR86-2 7.66 37.2 111.6 272.0 56.00 16.930 14.250
DR86-3 6.98 47.5 72.0 212.0 43.00 11.740 9.383
DR86-4 7.02 53.6 67.2 252.0 12.20 9.871 8.648
DR86-5 6.29 60.7 50.8 200.0 10.10 8.289 7.648
DR115-1 7.97 60.6 221.2 268.0 84.30 29.120 22.900
DR115-2 7.52 81.2 65.2 260.0 72.00 11.220 9.484
DR115-3 7.44 83.6 39.6 196.0 36.00 8.178 7.324
DR115-4 7.37 76.6 29.2 88.0 10.90 6.798 6.268
DR135-1 6.59 102.5 94.4 564.0 42.00 19.100 17.130
DR135-2 7.79 108.0 30.4 384.0 23.00 9.077 8.269
DR135-3 7.70 95.2 29.2 365.0 16.60 6.869 7.123
DR135-4 7.34 81.0 16.8 276.0 12.50 6.069 7.044
272
Table 5 Contd:
Identification TNO2
- (mg/L)
DNO2-
(mg/L) TNO3
- (mg/L)
DNO3-
(mg/L) TKN
(mg/L) DKN
(mg/L) TN
(mg/L) DTN
(mg/L) DR20-1 0.019 0.019 0.681 0.593 7.089 6.916 7.789 7.527
DR20-2 0.013 0.008 0.308 0.268 6.861 5.024 7.182 5.300
DR20-3 0.009 0.006 0.120 0.112 4.735 4.131 4.864 4.249
DR20-4 0.004 <0.001 0.111 0.110 3.289 2.718 3.404 2.828
DR20-5 0.003 <0.001 0.108 0.062 2.771 2.659 2.881 2.721
DR20-6 0.001 <0.001 0.097 0.050 1.597 1.403 1.695 1.453
DR20-7 0.002 <0.001 0.079 0.058 0.849 0.690 0.930 0.748
DR20-8 0.001 <0.001 0.083 0.052 0.711 0.530 0.795 0.582
DR40-1 <0.001 <0.001 0.450 0.360 10.546 6.098 10.995 6.458
DR40-2 <0.001 <0.001 0.156 0.108 5.224 4.627 5.380 4.735
DR40-3 <0.001 <0.001 0.092 0.078 4.364 3.442 4.456 3.520
DR40-4 <0.001 <0.001 0.102 0.054 2.621 2.444 2.724 2.498
DR40-5 <0.001 <0.001 0.082 0.055 1.917 1.670 2.000 1.725
DR40-6 <0.001 <0.001 0.092 0.057 1.650 1.020 1.743 1.077
DR40-7 <0.001 <0.001 0.094 0.050 1.365 1.288 1.459 1.339
DR65-1 <0.001 <0.001 0.240 0.235 6.671 4.956 6.911 5.192
DR65-2 <0.001 <0.001 0.077 0.068 2.827 2.783 2.903 2.851
DR65-3 <0.001 <0.001 0.072 0.070 1.312 1.302 1.384 1.372
DR65-4 <0.001 <0.001 0.095 0.069 2.166 2.083 2.261 2.153
DR65-5 <0.001 <0.001 0.013 0.013 1.945 1.213 1.959 1.225
DR65-6 <0.001 <0.001 0.009 0.009 0.936 0.732 0.945 0.741
DR86-1 <0.001 <0.001 0.212 0.155 4.750 2.918 4.962 3.072
DR86-2 <0.001 <0.001 0.079 0.052 2.150 2.019 2.229 2.070
DR86-3 <0.001 <0.001 0.074 0.046 1.827 1.298 1.901 1.344
DR86-4 <0.001 <0.001 0.066 0.045 1.029 0.865 1.095 0.909
DR86-5 <0.001 <0.001 0.060 0.042 0.903 0.561 0.963 0.603
DR115-1 <0.001 <0.001 0.145 0.126 4.234 2.764 4.379 2.890
DR115-2 <0.001 <0.001 0.083 0.052 1.192 0.853 1.276 0.905
DR115-3 <0.001 <0.001 0.062 0.057 0.639 0.545 0.701 0.602
DR115-4 <0.001 <0.001 0.053 0.050 0.417 0.310 0.470 0.360
DR135-1 <0.001 <0.001 0.133 0.118 1.682 1.559 1.815 1.677
DR135-2 <0.001 <0.001 0.076 0.069 0.723 0.580 0.798 0.649
DR135-3 <0.001 <0.001 0.072 0.052 0.356 0.318 0.428 0.370
DR135-4 <0.001 <0.001 0.070 0.043 0.283 0.167 0.353 0.210
273
Table 5 Contd:
Identification TPO43-
(mg/L) DPO4
3-
(mg/L) TP
(mg/L) DTP
(mg/L) DR20-1 0.997 0.489 1.755 0.515
DR20-2 0.942 0.463 1.685 0.472
DR20-3 0.693 0.308 0.875 0.336
DR20-4 0.546 0.260 0.640 0.268
DR20-5 0.500 0.229 0.519 0.267
DR20-6 0.448 0.214 0.493 0.220
DR20-7 0.387 0.185 0.396 0.194
DR20-8 0.225 0.103 0.267 0.161
DR40-1 1.917 0.362 1.960 0.551
DR40-2 0.822 0.343 0.928 0.409
DR40-3 0.812 0.334 0.921 0.357
DR40-4 0.600 0.289 0.795 0.325
DR40-5 0.576 0.272 0.680 0.300
DR40-6 0.318 0.184 0.391 0.215
DR40-7 0.301 0.135 0.379 0.191
DR65-1 0.384 0.173 0.414 0.201
DR65-2 0.335 0.136 0.353 0.156
DR65-3 0.213 0.053 0.233 0.064
DR65-4 0.201 0.047 0.213 0.049
DR65-5 0.194 0.040 0.199 0.046
DR65-6 0.124 0.038 0.166 0.044
DR86-1 0.335 0.254 0.526 0.264
DR86-2 0.303 0.169 0.501 0.234
DR86-3 0.297 0.167 0.439 0.209
DR86-4 0.286 0.107 0.362 0.111
DR86-5 0.195 0.077 0.197 0.085
DR115-1 0.736 0.366 0.898 0.420
DR115-2 0.224 0.108 0.305 0.139
DR115-3 0.114 <0.03 0.257 0.114
DR115-4 0.045 <0.03 0.063 0.012
DR135-1 0.790 0.370 0.834 0.405
DR135-2 0.128 <0.03 0.217 0.094
DR135-3 0.120 0.086 0.160 0.095
DR135-4 0.090 0.024 0.099 0.038 Note: DR- Drumbeat Street; TTU- Turbidity; EC- Electrical conductivity; TNO2
-- Total nitrite-
nitrogen; DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic
carbon; TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl
nitrogen; DKN-Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS-
Total suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates;
DPO43-- Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
274
Table 6- Wash-off test results- Ceil Circuit site
Sample name Identification Particle size distribution Intensity and
duration <1 µm 1-75 µm 75-150
µm 150-
300 µm >300 µm
CE-20-0-5 CE20-1 0 25.41 9.64 8.90 56.06
CE-20-5-10 CE20-2 0.58 23.90 9.95 7.50 58.07
CE-20-10-15 CE20-3 0.48 16.67 5.30 3.17 74.38
CE-20-15-20 CE20-4 0.50 16.85 5.27 3.05 74.32
CE-20-20-25 CE20-5 1.41 13.24 4.72 2.60 78.03
CE-20-25-30 CE20-6 1.09 11.56 4.38 2.25 80.72
CE-20-30-35 CE20-7 0.79 10.49 4.36 2.33 82.03
CE-20-35-40 CE20-8 0.80 9.96 4.62 2.14 82.48
CE-40-0-5 CE40-1 0.83 36.35 10.66 7.61 44.54
CE-40-5-10 CE40-2 0 13.48 3.33 0.24 82.96
CE-40-10-15 CE40-3 0.81 33.55 8.12 0.80 56.73
CE-40-15-20 CE40-4 0.46 7.61 1.96 0.33 89.64
CE-40-20-25 CE40-5 0.40 5.10 3.36 0.37 90.78
CE-40-25-30 CE40-6 2.07 41.20 17.77 3.14 35.83
CE-40-30-35 CE40-7 0.56 17.64 8.88 1.79 71.14
CE-65-0-5 CE65-1 0.7 36.24 18.07 9.55 35.44
CE-65-5-10 CE65-2 0.52 32.98 16.83 9.81 39.86
CE-65-10-15 CE65-3 0.36 22.16 8.57 2.48 66.43
CE-65-15-20 CE65-4 0.36 22.43 6.76 1.95 68.51
CE-65-20-25 CE65-5 0.16 10.59 8.96 1.92 78.37
CE-65-25-30 CE65-6 0.15 24.27 8.27 0.10 67.20
CE-86-0-5 CE86-1 1.59 51.02 15.84 8.92 22.63
CE-86-5-10 CE86-2 0.31 21.23 10.72 6.05 61.68
CE-86-10-15 CE86-3 0.42 7.59 3.69 0.36 87.94
CE-86-15-20 CE86-4 0 11.24 4.92 5.71 78.12
CE-86-20-25 CE86-5 0.23 10.87 5.88 1.31 81.72
CE-115-0-5 CE115-1 1.22 52.05 13.25 4.34 29.14
CE-115-5-10 CE115-2 1.26 55.21 16.77 5.05 21.70
CE-115-10-15 CE115-3 0 7.25 1.11 0.02 91.63
CE-115-15-20 CE115-4 0.28 7.42 2.28 0.10 89.92
CE-135-0-5 CE135-1 1.07 44.93 11.04 7.63 35.34
CE-135-5-10 CE135-2 0.84 36.88 12.04 5.06 45.20
CE-135-10-15 CE135-3 0.00 6.55 0.58 0.01 92.86
CE-135-15-20 CE135-4 1.36 55.67 25.60 3.48 13.88
275
Table 6 Contd:
Identification pH EC (µS/cm)
TSS (mg/L)
TDS (mg/L)
TTU (NTU)
TOC (mg/L)
DOC (mg/L)
CE20-1 6.69 66.6 95.2 192.0 37.0 45.430 38.860
CE20-2 6.99 39.6 29.6 132.0 9.6 34.370 31.930
CE20-3 7.08 34.7 18.2 94.0 8.8 25.800 25.330
CE20-4 7.16 34.0 15.6 86.0 6.6 18.030 15.230
CE20-5 8.16 34.3 11.8 72.0 3.8 15.660 13.770
CE20-6 8.01 34.8 10.4 60.0 3.4 12.410 11.750
CE20-7 7.91 34.7 9.2 56.0 2.9 10.380 9.890
CE20-8 7.50 32.6 8.0 44.0 2.4 8.978 8.684
CE40-1 7.07 57.5 101.2 152.0 27.0 33.870 29.450
CE40-2 7.18 40.3 27.6 106.0 16.8 18.270 17.200
CE40-3 7.17 38.6 13.6 98.0 7.7 12.890 11.990
CE40-4 7.15 35.4 11.6 78.0 6.3 8.972 8.346
CE40-5 7.23 34.5 10.8 72.0 6.1 7.212 6.783
CE40-6 7.31 32.8 7.4 66.0 5.0 6.123 5.673
CE40-7 7.30 33.1 6.8 52.0 2.6 5.459 5.237
CE65-1 6.33 102.4 74.4 162.0 13.6 19.460 16.450
CE65-2 6.59 74.6 26.0 152.0 8.4 9.580 9.162
CE65-3 6.70 78.3 18.0 148.0 7.6 7.784 7.426
CE65-4 6.74 70.8 13.2 128.0 7.3 6.999 6.832
CE65-5 6.77 71.2 9.6 110.0 6.4 6.143 5.922
CE65-6 6.85 76.4 8.4 94.0 4.7 5.601 5.507
CE86-1 6.93 86.4 42.4 144.0 8.9 14.390 12.960
CE86-2 6.94 74.9 12.4 134.0 7.0 9.442 8.782
CE86-3 6.99 84.1 11.6 128.0 6.6 8.759 8.378
CE86-4 7.12 83.9 7.6 96.0 5.9 7.799 7.297
CE86-5 7.15 81.2 6.4 62.0 4.8 6.577 6.307
CE115-1 6.93 89.4 55.2 164.0 8.8 17.300 15.480
CE115-2 7.01 59.7 32.4 110.0 5.9 10.270 9.863
CE115-3 6.99 60.4 13.6 100.0 5.0 8.293 7.847
CE115-4 6.97 59.4 8.0 92.0 3.3 7.152 6.686
CE135-1 6.96 75.4 74.4 142.0 8.2 17.650 15.830
CE135-2 7.06 63.5 17.6 114.0 5.4 9.880 9.222
CE135-3 7.03 49.4 9.2 92.0 4.0 7.328 6.498
CE135-4 7.32 44.2 7.6 74.0 2.9 5.447 5.407
276
Table 6 Contd:
Identification TNO2-
(mg/L) DNO2
- (mg/L)
TNO3-
(mg/L) DNO3
- (mg/L)
TKN (mg/L)
DKN (mg/L)
TN (mg/L)
DTN (mg/L)
CE20-1 <0.001 <0.001 0.414 0.332 10.160 7.818 10.574 8.150
CE20-2 <0.001 <0.001 0.183 0.117 7.341 5.852 7.524 5.969
CE20-3 <0.001 <0.001 0.168 0.103 6.072 4.262 6.240 4.365
CE20-4 <0.001 <0.001 0.160 0.102 4.321 3.159 4.481 3.261
CE20-5 <0.001 <0.001 0.154 0.101 2.637 2.601 2.790 2.702
CE20-6 <0.001 <0.001 0.148 0.094 2.543 2.135 2.691 2.229
CE20-7 <0.001 <0.001 0.147 0.084 2.489 1.784 2.636 1.868
CE20-8 <0.001 <0.001 0.138 0.072 2.462 1.537 2.600 1.609
CE40-1 <0.001 <0.001 0.221 0.155 8.723 5.772 8.944 5.927
CE40-2 <0.001 <0.001 0.184 0.105 3.923 2.878 4.107 2.983
CE40-3 <0.001 <0.001 0.172 0.100 2.797 2.061 2.969 2.161
CE40-4 <0.001 <0.001 0.167 0.093 2.054 1.296 2.221 1.389
CE40-5 <0.001 <0.001 0.161 0.085 1.719 1.050 1.879 1.135
CE40-6 <0.001 <0.001 0.145 0.082 1.527 0.833 1.672 0.914
CE40-7 <0.001 <0.001 0.137 0.742 1.428 0.767 1.565 1.509
CE65-1 <0.001 <0.001 0.326 0.281 3.451 2.382 3.777 2.663
CE65-2 <0.001 <0.001 0.189 0.095 2.192 1.232 2.381 1.327
CE65-3 <0.001 <0.001 0.172 0.094 1.746 0.952 1.918 1.045
CE65-4 <0.001 <0.001 0.168 0.090 1.603 0.782 1.770 0.872
CE65-5 <0.001 <0.001 0.161 0.086 1.501 0.695 1.661 0.781
CE65-6 <0.001 <0.001 0.102 0.074 1.247 0.646 1.349 0.719
CE86-1 <0.001 <0.001 0.185 0.099 2.204 2.139 2.389 2.238
CE86-2 <0.001 <0.001 0.177 0.091 2.074 1.464 2.251 1.555
CE86-3 <0.001 <0.001 0.171 0.092 1.856 1.208 2.027 1.300
CE86-4 <0.001 <0.001 0.167 0.161 1.632 0.847 1.799 1.007
CE86-5 <0.001 <0.001 0.167 0.089 1.299 0.730 1.465 0.819
CE115-1 <0.001 <0.001 0.179 0.098 4.109 2.843 4.288 2.941
CE115-2 <0.001 <0.001 0.151 0.081 2.499 1.312 2.650 1.393
CE115-3 <0.001 <0.001 0.128 0.068 2.075 1.167 2.203 1.235
CE115-4 <0.001 <0.001 0.110 0.060 1.623 0.870 1.733 0.929
CE135-1 <0.001 <0.001 0.182 0.093 4.124 2.792 4.306 2.885
CE135-2 <0.001 <0.001 0.152 0.083 2.192 1.416 2.344 1.498
CE135-3 <0.001 <0.001 0.146 0.075 1.640 0.914 1.786 0.989
CE135-4 <0.001 <0.001 0.125 0.065 1.251 0.645 1.376 0.710
277
Table 6 Contd: Identification TPO 4
3-
(mg/L) DPO4
3-
(mg/L) TP (mg/L)
DTP (mg/L)
CE20-1 0.304 0.145 0.361 0.155
CE20-2 0.279 <0.03 0.291 0.104
CE20-3 0.167 <0.03 0.180 0.075
CE20-4 0.070 <0.03 0.091 0.040
CE20-5 0.062 <0.03 0.085 0.039
CE20-6 0.054 <0.03 0.073 0.026
CE20-7 0.037 <0.03 0.039 0.017
CE20-8 0.022 <0.03 0.038 0.011
CE40-1 0.406 0.216 0.447 0.198
CE40-2 0.359 0.159 0.401 0.185
CE40-3 0.286 0.088 0.236 0.104
CE40-4 0.210 <0.03 0.226 0.061
CE40-5 0.053 <0.03 0.081 0.049
CE40-6 0.042 <0.03 0.056 0.038
CE40-7 0.024 <0.03 0.048 0.035
CE65-1 0.462 0.199 0.530 0.233
CE65-2 0.366 0.107 0.453 0.155
CE65-3 0.281 0.104 0.370 0.114
CE65-4 0.163 0.098 0.178 0.099
CE65-5 0.101 0.028 0.138 0.039
CE65-6 0.086 0.011 0.102 0.028
CE86-1 0.394 0.192 0.428 0.204
CE86-2 0.255 0.115 0.291 0.163
CE86-3 0.143 0.058 0.180 0.076
CE86-4 0.103 0.050 0.112 0.075
CE86-5 0.047 0.017 0.078 0.024
CE115-1 0.331 0.150 0.372 0.159
CE115-2 0.299 0.138 0.300 0.154
CE115-3 0.236 0.107 0.255 0.114
CE115-4 0.198 0.085 0.235 0.105
CE135-1 0.359 0.182 0.379 0.199
CE135-2 0.340 0.167 0.365 0.188
CE135-3 0.235 0.130 0.270 0.133
CE135-4 0.105 0.042 0.113 0.048 Note: CE- Ceil Circuit; TTU- Turbidity; EC- Electrical conductivity; TNO2
-- Total nitrite-nitrogen;
DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon;
TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen;
DKN- Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total
suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3--
Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
278
Table 7- Wash-off test results – Roof surfaces ; T-Tile roof; S- Steel roof
Sample name Sample identificatio
n
Particle size distribution (%)
Intensity and duration (min) <1 µm 1-75 µm 75-150 µm
150-300 µm
>300 µm
S-20-(0-1.95) S20-1 1.48 56.26 17.44 12.79 12.02
S-20-(1.95-3.90) S20-2 1.37 55.52 7.33 1.36 34.42
S-20-(3.90-5.62) S20-3 1.97 84.29 5.53 1.67 6.54
S-20-(5.62-7.70) S20-4 0.90 17.72 0.77 0.59 80.02
S-20-(7.70-10) S20-5 1.13 18.49 2.65 0.26 77.46
S-20-(10-11.83) S20-6 0 17.23 3.41 1.54 77.81
T-40-(0-1.50) T40-1 2.04 68.72 18.87 8.02 2.35
T-40-(1.50-2.42) T40-2 2.14 66.29 19.21 7.67 4.68
T-40-(2.42-3.50) T40-3 2.08 59.53 16.98 9.01 12.39
T-40-(3.50-4.48) T40-4 1.16 41.98 14.68 7.22 34.96
T-40-(4.48-5.42) T40-5 1.18 35.11 13.31 9.14 41.25
T-40-(5.42-6.33) T40-6 0 82.89 17.11 0.00 0
T-65-(0-0.25 ) T65-1 1.43 62.2 15.67 7.6 13.1
T-65-(0.25-0.50 ) T65-2 0.06 9.7 2.47 0.43 87.35
T-65-(0.50-0.75 ) T65-3 0.00 0 0 0 100
T-65-(0.75-1) T65-4 0.05 12.75 0 0 87.2
T-65-(1-1.25) T65-5 0.00 0 0 0 100
T-65-(1.25-1.50) T65-6 0.00 0 0 0 100
S-86-(0-1) S86-1 1.09 60.36 5.63 6.39 26.53
S-86-(1-1.50) S86-2 0.43 99.57 0.00 0 0.00
S-86(1.50, 2) S86-3 0 100 0.00 0.00 0.00
S-86-(2-2.67) S86-4 0 100 0.00 0.00 0.00
S-86-(2.67-3.17) S86-5 0 100 0.00 0.00 0.00
S-86-(3.17-3.67) S86-6 0 44.27 1.94 3.82 49.97
T-115-(0-0.50) T115-1 0.89 40.32 13.09 3.7 41.99
T-115-(0.50-0.83) T115-2 1.07 37.83 6.90 1.62 52.58
T-115-(0.83-1.07) T115-3 0.96 39.84 7.46 1.17 50.57
T-115-(1.07-1.33) T115-4 0.00 7.31 0.28 0.00 92.41
T-115-(1.33-1.67) T115-5 0.79 17.17 2.37 0.05 79.62
S-135-(0-0.33) S135-1 1.68 61.49 13.10 10.51 13.22
S-135-(0.33-0.53) S135-2 0.78 31.22 1.98 0.54 65.48
S-135-(0.53-0.75) S135-3 0.93 28.74 0.17 0.00 70.16
S-135-(0.75-1.17) S135-4 1.67 41.60 0.00 0.24 56.48
S-135-(1.17-1.50) S135-5 0.00 6.12 0.00 0.00 93.88
279
Table 7 Contd:
Sample identification pH
EC (µS/cm)
TTU (NTU)
TSS (mg/L)
TDS (mg/L)
TOC (mg/L)
DOC (mg/L)
S20-1 6.36 80.90 1.1 84.4 54.0 4.929 4.162
S20-2 6.66 69.30 0.9 19.2 38.0 2.278 2.770
S20-3 6.75 54.52 0.6 16.0 28.0 2.124 2.263
S20-4 6.80 56.81 0.4 8.0 26.0 1.837 1.892
S20-5 7.31 62.70 0.3 7.2 24.0 2.747 2.562
S20-6 8.26 51.76 0.2 3.6 22.0 2.437 2.541
T40-1 6.98 82.70 2.2 50.8 100.0 3.650 3.084
T40-2 7.20 64.50 0.8 14.8 86.0 1.780 2.917
T40-3 7.34 53.60 1.0 8.0 34.0 1.679 2.859
T40-4 7.31 55.31 1.0 6.8 26.0 1.222 2.213
T40-5 7.30 55.97 0.7 4.4 24.0 1.077 1.376
T40-6 7.34 55.38 0.9 3.6 20.0 1.079 1.696
T65-1 6.51 59.68 10.1 55.2 96.0 5.936 4.659
T65-2 6.82 38.52 8.2 10.8 84.0 3.362 2.264
T65-3 6.83 38.01 6.2 7.2 72.0 2.374 2.082
T65-4 6.86 36.28 5.0 6.0 68.0 2.390 2.190
T65-5 6.85 35.59 4.8 5.2 54.0 2.381 2.704
T65-6 6.84 35.84 4.6 2.8 32.0 2.420 8.352
S86-1 6.99 86.22 11.5 71.2 66.0 11.080 2.800
S86-2 7.54 64.98 9.5 20.0 54.0 3.589 2.620
S86-3 7.65 55.4 8.4 5.6 50.0 2.691 2.583
S86-4 7.58 55.09 5.5 5.2 30.0 2.908 3.110
S86-5 8.20 55.19 5.3 4.0 22.0 2.601 3.010
S86-6 8.38 45.60 5.0 3.6 18.0 6.420 4.779
T115-1 7.02 83.00 6.8 86.8 114.0 4.096 5.807
T115-2 7.22 72.00 6.2 15.6 58.0 2.996 6.576
T115-3 7.29 67.60 4.4 13.6 50.0 2.721 3.642
T115-4 7.28 69.60 2.5 11.6 48.0 2.626 2.727
T115-5 7.21 63.60 2.1 9.2 34.0 2.476 3.292
S135-1 7.96 79.70 4.9 45.6 54.0 5.683 4.511
S135-2 8.26 67.30 1.2 11.6 46.0 2.904 3.499
S135-3 8.30 67.70 1.6 6.4 30.0 2.425 3.172
S135-4 8.32 67.90 0.8 3.6 26.0 2.451 2.952
S135-5 8.31 68.20 0.8 3.2 22.0 2.474 3.084
280
Table 7 Contd:
Sample identification
TNO2-
(mg/L) DNO2
- (mg/L)
TNO3- (mg/L)
DNO3-
(mg/L) TKN
(mg/L) DKN
(mg/L) TN
(mg/L) DTN
(mg/L) S20-1 0.029 0.025 0.107 0.098 0.658 0.580 0.793 0.703
S20-2 0.032 0.020 0.103 0.081 0.578 0.477 0.712 0.578
S20-3 0.032 0.018 0.093 0.072 0.458 0.387 0.582 0.477
S20-4 0.023 0.017 0.090 0.063 0.426 0.343 0.538 0.423
S20-5 0.025 0.014 0.082 0.058 0.390 0.339 0.497 0.410
S20-6 0.026 0.013 0.072 0.043 0.307 0.260 0.405 0.316
T40-1 0.035 0.029 0.106 0.099 0.841 0.666 0.982 0.794
T40-2 0.030 0.026 0.096 0.090 0.613 0.604 0.739 0.720
T40-3 0.031 0.020 0.095 0.070 0.423 0.320 0.549 0.410
T40-4 0.025 0.019 0.082 0.051 0.323 0.313 0.429 0.382
T40-5 0.030 0.018 0.078 0.043 0.259 0.246 0.368 0.307
T40-6 0.029 0.016 0.070 0.036 0.225 0.213 0.324 0.266
T65-1 0.037 0.032 0.253 0.239 1.374 0.890 1.664 1.160
T65-2 0.036 0.029 0.112 0.096 0.911 0.848 1.059 0.973
T65-3 0.034 0.022 0.111 0.082 0.895 0.711 1.040 0.815
T65-4 0.031 0.020 0.102 0.071 0.652 0.439 0.785 0.531
T65-5 0.032 0.019 0.093 0.063 0.463 0.376 0.588 0.457
T65-6 0.032 0.017 0.090 0.056 0.317 0.246 0.439 0.319
S86-1 0.039 0.030 0.496 0.351 1.177 0.796 1.712 1.177
S86-2 0.037 0.025 0.259 0.213 0.555 0.506 0.851 0.744
S86-3 0.036 0.020 0.225 0.211 0.532 0.505 0.793 0.736
S86-4 0.035 0.019 0.212 0.204 0.472 0.436 0.719 0.659
S86-5 0.034 0.018 0.205 0.195 0.468 0.426 0.707 0.639
S86-6 0.030 0.016 0.108 0.105 0.372 0.350 0.509 0.472
T115-1 0.034 0.029 0.107 0.103 0.649 0.569 0.790 0.701
T115-2 0.031 0.025 0.104 0.097 0.553 0.475 0.688 0.597
T115-3 0.030 0.021 0.103 0.096 0.482 0.455 0.614 0.572
T115-4 0.031 0.019 0.088 0.054 0.388 0.310 0.506 0.383
T115-5 0.028 0.016 0.064 0.048 0.316 0.309 0.409 0.374
S135-1 0.039 0.030 0.186 0.147 0.784 0.520 1.008 0.697
S135-2 0.036 0.024 0.151 0.129 0.539 0.414 0.727 0.567
S135-3 0.027 0.018 0.075 0.051 0.475 0.324 0.577 0.393
S135-4 0.026 0.015 0.096 0.047 0.428 0.315 0.549 0.377
S135-5 0.024 0.013 0.083 0.043 0.310 0.299 0.417 0.356
281
Table 7 Contd:
Sample identification
TPO43-
(mg/L) DPO4
3-
(mg/L) TP
(mg/L) DTP
(mg/L) S20-1 4.275 2.693 4.495 1.739
S20-2 4.149 1.518 4.250 1.698
S20-3 4.074 1.352 4.284 1.471
S20-4 3.594 1.274 3.433 0.354
S20-5 1.422 0.184 2.525 0.283
S20-6 0.915 0.108 1.322 0.466
T40-1 2.449 0.370 2.961 1.276
T40-2 2.258 0.316 2.578 1.079
T40-3 1.125 0.215 1.145 0.472
T40-4 1.109 0.210 1.139 0.350
T40-5 0.576 0.153 0.627 0.235
T40-6 0.365 0.128 0.521 0.159
T65-1 0.231 0.110 0.252 0.187
T65-2 0.015 0.015 0.166 0.080
T65-3 0.015 0.015 0.114 0.057
T65-4 0.015 0.015 0.098 0.046
T65-5 0.015 0.015 0.094 0.035
T65-6 0.015 0.015 0.087 0.018
S86-1 0.250 0.085 0.372 0.109
S86-2 0.078 0.038 0.090 0.043
S86-3 0.043 0.015 0.074 0.022
S86-4 0.015 0.015 0.059 0.021
S86-5 0.015 0.015 0.038 0.011
S86-6 0.015 0.015 0.012 0.008
T115-1 3.313 0.670 4.230 2.204
T115-2 3.050 0.667 4.588 2.086
T115-3 2.150 0.666 3.025 1.170
T115-4 1.160 0.444 2.322 1.105
T115-5 0.985 0.364 2.090 0.099
S135-1 4.599 0.986 5.327 1.456
S135-2 3.542 0.950 3.123 1.432
S135-3 1.098 0.602 2.566 1.206
S135-4 0.877 0.243 1.065 0.366
S135-5 0.743 0.164 1.033 0.271 Note: S- Steel roof; T-Tile roof; TTU- Turbidity; EC- Electrical conductivity; TNO2
-- Total nitrite-
nitrogen; DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic
carbon; TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl
nitrogen; DKN Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS-
Total suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates;
DPO43-- Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
282
Appendix C
Build-up Analysis
283
Appendix C
Build-up Analysis
284
Appendix C
Build-up Analysis
285
Table 1- Amounts of pollutants in different particle size fractions of build- up (mg/g)- road surfaces
Particle size class
(µm)
TS
TOC
NO2-
NO3
-
TKN
TN
PO43-
TP
DR<1 180.79 12.25 0.00 0.70 1.07 1.77 0.04 0.10
DR-1-75 415.82 17.68 0.00 0.02 2.03 2.05 1.58 1.59
DR-75-150 194.35 8.26 0.00 0.10 1.90 1.99 1.44 1.46
DR-150-300 109.04 3.93 0.00 0.12 0.09 0.21 0.97 1.00
DR>300 100.00 3.56 0.00 0.11 0.61 0.72 0.53 0.54
CE<1 329.78 8.74 0.00 0.58 0.23 0.81 0.28 0.30
CE-1-75 14.21 9.51 0.00 0.25 0.61 0.86 2.49 2.65
CE-75-150 507.86 17.21 0.00 0.13 4.07 4.20 2.08 2.43
CE-150-300 54.79 10.25 0.00 0.07 0.05 0.13 1.39 1.98
CE>300 93.35 5.06 0.00 0.07 0.20 0.26 0.91 2.26 Note; DR- Drumbeat Street; CE- Ceil Circuit; TS- Total solids; TOC- Total organic carbon; NO2
--
nitrite-nitrogen; NO3-- nitrate- nitrogen; TKN- Total kjeldahl nitrogen; TN- Total nitrogen; TPO4
3--
Total Phosphates; TP- Total phosphorus.
286
Table 2- Average build-up data for roof surfaces
Particle size distribution (%) Sampling episode
Sample name
Volume (L)
Antecedent dry days <1 µm 1-75 µm 75-150 µm
150-300 µm >300 µm
1 BU 1 6.40 8 1.25 65.59 10.78 6.51 15.88 2 BU 2 6.39 6 1.76 66.89 11.04 4.10 16.23 3 BU 3 7.07 6 3.72 60.59 16.57 8.60 10.53
Table 2 Contd:
Sample name
TSS (mg/L)
TTU (NTU)
TOC (mg/L)
NO2-
(mg/L) NO3
- (mg/L)
TKN (mg/L)
TN (mg/L)
TPO43-
(mg/L) TP (mg/L)
BU 1 50.2 10.5 0.995 0.1440 0.3012 1.2489 1.6941 2.7100 2.9277
BU 2 41 8.55 3.932 0.2345 0.4055 1.9231 2.5630 3.3937 3.6603 BU 3 37.2 4.4 2.217 0.1285 0.2478 0.4931 0.8693 0.4118 0.6996
Note: TSS- Total suspended solids; TOC- Total organic carbon; TTU- Turbidity NO2-- nitrite-nitrogen; NO3
-- nitrate- nitrogen; TKN- Total kjeldahl nitrogen; TN-
Total nitrogen; TPO43-- Total Phosphates; TP- Total phosphorus.
287
Table 3- Amount of pollutants in different particle size fractions (mg/g) –Roof surfaces
Particle size class (µm)
TS (mg/g)
TOC (mg/g)
NO2- (mg/g)
NO3- (mg/g)
TKN (mg/g)
TN (mg/g)
PO43-
(mg/g) TP (mg/g)
BU1<1 616.88 62.69 1.42 3.11 3.43 7.96 0.92 1.09
BU1-1-75 74.68 14.86 0.06 0.18 0.97 1.21 2.06 2.14
BU1-75-150 259.74 79.21 0.36 0.91 7.11 8.38 0.42 0.50
BU1-150-300 48.70 8.12 0.38 0.92 1.36 2.66 1.52 1.84
BU1>300 58.44 7.43 0.42 0.73 1.67 2.82 1.36 1.59
BU2<1 474.36 80.68 2.63 4.52 7.39 14.54 0.98 1.21
BU2-1-75 125.64 31.55 0.05 0.08 1.66 1.79 5.23 6.82
BU2-75-150 223.08 90.40 0.78 1.08 10.27 12.13 3.05 3.17
BU2-150-300 97.44 18.84 0.28 1.02 1.73 3.03 3.33 3.41
BU2>300 79.49 11.27 0.22 0.90 2.21 3.33 3.10 3.43
BU3<1 469.08 22.62 1.59 3.53 1.62 6.74 0.81 1.13
BU3-75-1 114.09 4.18 0.08 0.29 1.48 1.85 2.39 2.57
BU3-150-75 268.46 9.58 0.65 1.26 1.47 3.38 1.66 1.86
BU3-300-150 111.86 6.89 0.21 0.55 1.69 2.45 2.22 2.54
BU3>300 35.79 7.35 0.25 0.74 1.16 2.15 1.88 2.00 Note: TS- Total solids; TOC- Total organic carbon; NO2
-- nitrite-nitrogen; NO3-- nitrate- nitrogen;
TKN- Total kjeldahl nitrogen; TN- Total nitrogen; TPO43-- Total Phosphates; TP- Total phosphorus.
288
289
Appendix D
Wash-off Analysis- Data Matrices
290
291
Data matrices used for the wash-off analysis Table 1- Data matrix-Drumbeat Street site
Identification EC (µS/cm)
TTU (NTU)
TSS (mg/L/g)
TDS (mg/L/g)
TOC (mg/L/g)
DOC (mg/L/g)
D20-1 170.5 55 8.32 67.82 757.947 671.089
D20-2 107 37 5.96 56.00 428.437 373.762
D20-3 59.6 17.7 3.24 31.34 205.670 185.784
D20-4 52.1 14.4 2.77 24.15 156.709 136.961
D20-5 52.3 12.3 2.08 15.93 118.949 117.486
D20-6 50.9 11.4 1.90 8.73 98.698 85.715
D20-7 49.6 9.3 1.75 4.11 63.497 61.943
D20-8 50.3 6.7 1.54 2.31 71.818 70.766
D40-1 84.5 49 7.14 32.37 289.053 223.178
D40-2 47.9 30 4.93 28.51 143.178 117.029
D40-3 42.6 22 4.11 28.26 110.492 88.595
D40-4 42.2 19 3.29 25.69 81.646 72.412
D40-5 48.6 13 2.21 25.69 60.755 55.909
D40-6 45.3 10.5 1.59 24.15 55.315 52.206
D40-7 44.6 8.8 1.31 11.30 50.926 45.897
D65-1 114 19.8 16.90 16.44 145.098 121.235
D65-2 88.1 12.6 9.25 14.39 78.675 66.560
D65-3 121.3 11.8 6.47 11.56 55.315 52.435
D65-4 41 11.5 5.81 10.79 44.860 42.409
D65-5 53.9 10.4 4.52 9.50 42.277 38.798
D65-6 65 8.6 4.01 6.17 34.972 37.376
D86-1 48.3 94 28.51 42.64 135.726 104.412
D86-2 37.2 56 14.33 34.94 77.395 65.143
D86-3 47.5 43 9.25 27.23 53.669 42.894
D86-4 53.6 12.2 8.63 32.37 45.125 39.534
D86-5 60.7 10.1 6.53 25.69 37.893 34.962
D115-1 60.6 84.3 28.41 34.42 133.121 104.686
D115-2 81.2 72 8.37 33.40 51.292 43.356
D115-3 83.6 36 5.09 25.18 37.385 33.481
D115-4 76.6 10.9 3.75 11.30 31.077 28.654
D135-1 102.5 42 12.13 72.44 87.315 78.309
D135-2 108 23 3.90 49.32 41.495 37.801
D135-3 95.2 16.6 3.75 46.88 31.401 32.562
D135-4 81 12.5 2.16 35.45 27.744 32.201
292
Table 1 Contd:
Identification TNO2-
(mg/L/g) DNO2
- (mg/L/g)
TNO3-
(mg/L/g) DNO3
- (mg/L/g)
TKN (mg/L/g)
DKN (mg/L/g)
TN (mg/L/g)
DTN (mg/L/g)
D20-1 200.931 200.931 119.714 104.156 111.384 108.661 112.189 108.418
D20-2 137.479 84.602 54.161 47.112 107.796 78.938 103.441 76.338
D20-3 95.178 63.452 21.007 19.741 74.400 64.910 70.054 61.207
D20-4 42.301 5.288 19.478 19.249 51.685 42.711 49.033 40.731
D20-5 31.726 5.288 18.898 10.934 43.538 41.775 41.503 39.191
D20-6 10.575 5.288 17.087 8.807 25.087 22.043 24.411 20.929
D20-7 21.151 5.288 13.905 10.196 13.337 10.840 13.394 10.773
D20-8 10.575 5.288 14.555 9.124 11.175 8.331 11.451 8.384
D40-1 5.288 5.288 79.053 63.267 165.696 95.815 158.371 93.017
D40-2 5.288 5.288 27.459 19.021 82.079 72.694 77.492 68.197
D40-3 5.288 5.288 16.102 13.641 68.569 54.089 64.177 50.701
D40-4 5.288 5.288 17.983 9.510 41.187 38.393 39.230 35.975
D40-5 5.288 5.288 14.468 9.598 30.127 26.238 28.803 24.839
D40-6 5.288 5.288 16.243 9.950 25.932 16.030 25.103 15.510
D40-7 5.288 5.288 16.560 8.842 21.440 20.242 21.011 19.281
D65-1 5.288 5.288 42.260 41.346 104.815 77.877 99.547 74.778
D65-2 5.288 5.288 13.448 12.007 44.418 43.723 41.820 41.065
D65-3 5.288 5.288 12.587 12.341 20.616 20.461 19.930 19.768
D65-4 5.288 5.288 16.718 12.200 34.032 32.735 32.567 31.008
D65-5 5.288 5.288 2.356 2.215 30.567 19.053 28.214 17.647
D65-6 5.288 5.288 1.617 1.547 14.708 11.502 13.616 10.670
D86-1 5.288 5.288 37.268 27.177 74.627 45.841 71.464 44.250
D86-2 5.288 5.288 13.817 9.106 33.785 31.716 32.103 29.820
D86-3 5.288 5.288 13.026 8.157 28.710 20.393 27.386 19.363
D86-4 5.288 5.288 11.549 7.893 16.173 13.583 15.772 13.099
D86-5 5.288 5.288 10.460 7.454 14.190 8.808 13.865 8.685
D115-1 5.288 5.288 25.560 22.167 66.520 43.428 63.074 41.627
D115-2 5.288 5.288 14.661 9.106 18.732 13.400 18.373 13.030
D115-3 5.288 5.288 10.864 10.073 10.037 8.557 10.091 8.670
D115-4 5.288 5.288 9.335 8.807 6.555 4.863 6.774 5.180
D135-1 5.288 5.288 23.327 20.779 26.435 24.494 26.144 24.156
D135-2 5.288 5.288 13.272 12.200 11.359 9.105 11.500 9.347
D135-3 5.288 5.288 12.692 9.053 5.587 4.998 6.162 5.324
D135-4 5.288 5.288 12.376 7.594 4.445 2.618 5.089 3.022
293
Table 1 Contd:
Identification TPO43-
(mg/L/g) DPO4
3- (mg/L/)g
TP (mg/L/g)
DTP (mg/L/g)
D20-1 46.878 22.992 75.029 22.026
D20-2 44.292 21.770 72.049 20.196
D20-3 32.584 14.482 37.420 14.343
D20-4 25.672 12.225 27.352 11.470
D20-5 23.509 10.767 22.167 11.432
D20-6 21.064 10.062 21.059 9.418
D20-7 18.196 8.698 16.942 8.281
D20-8 10.579 4.843 11.415 6.862
D40-1 90.135 17.021 83.793 23.535
D40-2 38.649 16.127 39.661 17.494
D40-3 38.179 15.704 39.366 15.249
D40-4 28.211 13.588 33.992 13.873
D40-5 27.083 12.789 29.088 12.817
D40-6 14.952 8.651 16.703 9.183
D40-7 14.153 6.348 16.181 8.183
D65-1 18.055 8.134 17.712 8.576
D65-2 15.751 6.395 15.108 6.682
D65-3 10.015 2.492 9.940 2.745
D65-4 9.451 2.210 9.085 2.112
D65-5 9.122 1.881 8.499 1.945
D65-6 5.830 1.787 7.105 1.860
D86-1 15.751 11.943 22.487 11.265
D86-2 14.247 7.946 21.427 10.008
D86-3 13.965 7.852 18.776 8.935
D86-4 13.447 5.031 15.485 4.728
D86-5 9.169 3.620 8.409 3.651
D115-1 34.606 17.209 38.369 17.943
D115-2 10.532 5.078 13.039 5.951
D115-3 5.360 0.705 10.991 4.852
D115-4 2.116 0.705 2.681 0.492
D135-1 37.145 17.397 35.646 17.293
D135-2 6.018 0.705 9.260 4.010
D135-3 5.642 4.044 6.857 4.053
D135-4 4.232 1.128 4.228 1.633
Note: D- Drumbeat Street; TTU- Turbidity; EC- Electrical conductivity; TNO2-- Total nitrite-
nitrogen; DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic
carbon; TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl
nitrogen; DKN- Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen;
TSS- Total suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total
Phosphates; DPO43-- Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total
phosphorus.
294
Table 2- Ceil Circuit data matrix
Identification EC (µS/cm)
TTU (NTU)
TSS (mg/L/g)
TDS (mg/L/g)
TOC (mg/L/g)
DOC (mg/L/g)
C20-1 66.6 37.0 33.00 66.56 783.091 669.842
C20-2 39.6 9.6 10.26 45.76 592.446 550.387
C20-3 34.7 8.8 6.31 32.59 444.723 436.621
C20-4 34.0 6.6 5.41 29.82 310.789 262.524
C20-5 34.3 3.8 4.09 24.96 269.936 237.358
C20-6 34.8 3.4 3.61 20.80 213.915 202.538
C20-7 34.7 2.9 3.19 19.41 178.923 170.477
C20-8 32.6 2.4 2.77 15.25 154.757 149.689
C40-1 57.5 27.0 35.09 52.70 583.828 507.639
C40-2 40.3 16.8 9.57 36.75 314.926 296.482
C40-3 38.6 7.7 4.71 33.98 222.189 206.675
C40-4 35.4 6.3 4.02 27.04 154.653 143.863
C40-5 34.5 6.1 3.74 24.96 124.316 116.921
C40-6 32.8 5.0 2.57 22.88 105.544 97.787
C40-7 33.1 2.6 2.36 18.03 94.098 90.272
C65-1 102.4 13.6 25.79 56.16 335.438 283.554
C65-2 74.6 8.4 9.01 52.70 165.133 157.928
C65-3 78.3 7.6 6.24 51.31 134.175 128.004
C65-4 70.8 7.3 4.58 44.38 120.644 117.765
C65-5 71.2 6.4 3.33 38.14 105.889 102.079
C65-6 76.4 4.7 2.91 32.59 96.546 94.926
C86-1 86.4 8.9 14.70 49.92 248.045 223.396
C86-2 74.9 7.0 4.30 46.46 162.755 151.378
C86-3 84.1 6.6 4.02 44.38 150.982 144.414
C86-4 83.9 5.9 2.63 33.28 134.434 125.781
C86-5 81.2 4.8 2.22 21.49 113.370 108.716
C115-1 89.4 8.8 19.14 56.86 298.206 266.834
C115-2 59.7 5.9 11.23 38.14 177.027 170.012
C115-3 60.4 5.0 4.71 34.67 142.949 135.261
C115-4 59.4 3.3 2.77 31.90 123.281 115.249
C135-1 75.4 8.2 25.79 49.23 304.239 272.867
C135-2 63.5 5.4 6.10 39.52 170.305 158.963
C135-3 49.4 4.0 3.19 31.90 126.315 112.008
C135-4 44.2 2.9 2.63 25.66 93.892 93.202
295
Table 2 Contd:
Identification TNO2-
(mg/L/g) DNO2
- (mg/L/g)
TNO3-
(mg/L/g) DNO3
- (mg/L/g)
TKN (mg/L/g)
DKN (mg/L/g)
TN (mg/L/g)
DTN (mg/L/g)
C20-1 150.602 150.602 168.308 135.054 396.778 305.313 376.777 290.408
C20-2 150.602 150.602 74.446 47.460 286.685 228.551 268.105 212.698
C20-3 150.602 150.602 68.504 41.965 237.120 166.434 222.358 155.538
C20-4 150.602 150.602 65.125 41.395 168.738 123.371 159.668 116.194
C20-5 150.602 150.602 62.602 41.070 102.966 101.584 99.432 96.286
C20-6 150.602 150.602 60.322 38.424 99.303 83.358 95.890 79.424
C20-7 150.602 150.602 59.793 34.150 97.186 69.662 93.913 66.553
C20-8 150.602 150.602 56.008 29.306 96.151 60.024 92.637 57.335
C40-1 150.602 150.602 89.832 62.927 340.652 225.412 318.694 211.187
C40-2 150.602 150.602 74.813 42.820 153.219 112.397 146.355 106.306
C40-3 150.602 150.602 69.928 40.825 109.230 80.483 105.789 77.012
C40-4 150.602 150.602 67.852 37.895 80.206 50.620 79.125 49.506
C40-5 150.602 150.602 65.451 34.557 67.112 41.013 66.966 40.448
C40-6 150.602 150.602 58.938 33.173 59.625 32.519 59.565 32.576
C40-7 150.602 150.602 55.601 302.019 55.767 29.961 55.753 53.778
C65-1 150.602 150.602 132.693 114.417 134.755 93.008 134.574 94.882
C65-2 150.602 150.602 76.929 38.587 85.592 48.128 84.833 47.293
C65-3 150.602 150.602 69.969 38.058 68.170 37.174 68.328 37.252
C65-4 150.602 150.602 68.178 36.511 62.597 30.535 63.086 31.058
C65-5 150.602 150.602 65.410 34.923 58.606 27.130 59.202 27.812
C65-6 150.602 150.602 41.517 29.917 48.691 25.216 48.063 25.628
C86-1 150.602 150.602 75.220 40.296 86.068 83.533 85.118 79.748
C86-2 150.602 150.602 71.964 36.959 80.991 57.181 80.201 55.410
C86-3 150.602 150.602 69.725 37.488 72.486 47.176 72.244 46.327
C86-4 150.602 150.602 68.137 65.451 63.722 33.062 64.109 35.897
C86-5 150.602 150.602 67.812 36.348 50.718 28.504 52.214 29.191
C115-1 150.602 150.602 72.900 39.686 160.451 111.038 152.787 104.792
C115-2 150.602 150.602 61.462 32.970 97.589 51.245 94.426 49.645
C115-3 150.602 150.602 51.978 27.515 81.030 45.578 78.487 43.997
C115-4 150.602 150.602 44.937 24.218 63.382 33.960 61.768 33.107
C135-1 150.602 150.602 74.039 37.895 161.049 109.031 153.432 102.803
C135-2 150.602 150.602 61.706 33.580 85.619 55.283 83.525 53.383
C135-3 150.602 150.602 59.305 30.528 64.058 35.682 63.642 35.231
C135-4 150.602 150.602 51.001 26.294 48.855 25.205 49.043 25.300
296
Table 2 Contd:
Identification TPO4
3- (mg/L/g)
DPO43-
(mg/L/g) TP
(mg/L/g) DTP
(mg/L/g)
C20-1 64.122 30.585 66.145 28.293
C20-2 58.849 3.164 53.308 19.082
C20-3 35.225 3.164 32.944 13.679
C20-4 14.765 3.164 16.628 7.362
C20-5 13.078 3.164 15.639 7.215
C20-6 11.390 3.164 13.313 4.670
C20-7 7.804 3.164 7.142 3.131
C20-8 4.640 3.164 6.995 1.923
C40-1 85.637 45.560 81.857 36.204
C40-2 75.723 33.538 73.360 33.915
C40-3 60.325 18.562 43.291 19.118
C40-4 44.295 3.164 41.423 11.079
C40-5 11.179 3.164 14.778 8.900
C40-6 8.859 3.164 10.310 6.867
C40-7 5.062 3.164 8.717 6.483
C65-1 97.449 41.975 97.075 42.723
C65-2 77.200 22.569 82.883 28.348
C65-3 59.271 21.936 67.775 20.895
C65-4 34.381 20.671 32.523 18.129
C65-5 21.304 5.906 25.290 7.069
C65-6 18.140 2.320 18.697 5.128
C86-1 83.106 40.498 78.304 37.303
C86-2 53.787 24.257 53.308 29.776
C86-3 30.163 12.234 32.889 13.918
C86-4 21.726 10.546 20.583 13.661
C86-5 9.914 3.586 14.357 4.303
C115-1 69.817 31.639 68.178 29.154
C115-2 63.067 29.108 54.883 28.110
C115-3 49.779 22.569 46.697 20.876
C115-4 41.764 17.929 43.034 19.228
C135-1 75.723 38.389 69.405 36.369
C135-2 71.715 35.225 66.841 34.373
C135-3 49.568 27.421 49.371 24.264
C135-4 22.147 8.859 20.748 8.753
Note: C- Ceil Circuit; TTU- Turbidity; EC- Electrical conductivity; TNO2-- Total nitrite-nitrogen;
DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon;
TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen;
DKN- Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total
suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3--
Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
297
Table 3- Roof surfaces –data matrix
Identification EC (µS/cm)
TTU (NTU)
TSS (mg/L/g)
TDS (mg/L/g)
TOC (mg/L/g)
DOC (mg/L/g)
R20-1 80.90 1.10 154.63 98.94 314.537 265.592
R20-2 69.30 0.90 35.18 69.62 145.367 176.764
R20-3 54.52 0.60 29.31 51.30 135.540 144.410
R20-4 56.81 0.40 14.66 47.64 117.225 120.735
R20-5 62.70 0.30 13.19 43.97 175.296 163.490
R20-6 51.76 0.20 6.60 40.31 155.514 162.150
R40-1 82.70 2.20 93.07 183.22 232.919 196.801
R40-2 64.50 0.80 27.12 157.57 113.588 186.144
R40-3 53.60 1.00 14.66 62.29 107.143 182.443
R40-4 55.31 1.00 12.46 47.64 77.980 141.219
R40-5 55.97 0.70 8.06 43.97 68.727 87.807
R40-6 55.38 0.90 6.60 36.64 68.855 108.228
R65-1 59.68 10.10 95.62 166.30 931.739 731.296
R65-2 38.52 8.20 18.71 145.51 527.714 355.367
R65-3 38.01 6.20 12.47 124.72 372.633 326.799
R65-4 36.28 5.00 10.39 117.79 375.144 343.752
R65-5 35.59 4.80 9.01 93.54 373.732 424.431
R65-6 35.84 4.60 4.85 55.43 379.853 1310.965
R86-1 86.22 11.50 123.34 114.33 1739.163 439.500
R86-2 64.98 9.50 34.65 93.54 563.344 411.246
R86-3 55.40 8.40 9.70 86.61 422.391 405.438
R86-4 55.09 5.50 9.01 51.97 456.452 488.159
R86-5 55.19 5.30 6.93 38.11 408.264 472.462
R86-6 45.60 5.00 6.24 31.18 1007.710 750.132
R115-1 83.00 6.80 156.13 205.06 163.022 231.120
R115-2 72.00 6.20 28.06 104.33 119.242 261.726
R115-3 67.60 4.40 24.46 89.94 108.296 144.952
R115-4 69.60 2.50 20.87 86.34 104.515 108.535
R115-5 63.60 2.10 16.55 61.16 98.545 131.022
R135-1 79.70 4.90 82.02 97.13 226.185 179.539
R135-2 67.30 1.20 20.87 82.74 115.580 139.261
R135-3 67.70 1.60 11.51 53.96 96.516 126.246
R135-4 67.90 0.80 6.48 46.77 97.550 117.490
R135-5 68.20 0.80 5.76 39.57 98.466 122.744
298
Table 3 Contd:
Identification TNO2
- (mg/L/g)
DNO2-
(mg/L/g) TNO3
- (mg/L/g)
DNO3-
(mg/L/g) TKN
(mg/L/g) DKN
(mg/L/g) TN
(mg/L/g) DTN
(mg/L/g)
R20-1 31.921 27.518 60.973 55.835 188.619 166.415 129.077 114.368
R20-2 35.223 22.014 58.575 46.301 165.669 136.724 115.865 93.997
R20-3 35.223 19.813 52.866 40.934 131.273 111.135 94.729 77.628
R20-4 25.317 18.712 51.211 36.139 122.064 98.254 87.570 68.793
R20-5 27.518 15.410 46.643 32.999 111.852 97.135 80.801 66.776
R20-6 28.619 14.309 41.048 24.720 88.185 74.444 65.946 51.383
R40-1 38.525 31.921 60.745 56.577 241.117 191.057 159.764 129.207
R40-2 33.022 28.619 54.636 51.211 175.882 173.357 120.209 117.150
R40-3 34.122 22.014 54.293 39.849 121.204 91.885 89.262 66.727
R40-4 27.518 20.914 46.529 28.831 92.603 89.705 69.851 62.187
R40-5 33.022 19.813 44.702 24.606 74.415 70.657 59.828 50.017
R40-6 31.921 17.612 39.792 20.667 64.489 61.190 52.636 43.199
R65-1 40.148 34.722 131.090 123.931 171.914 111.285 153.455 107.026
R65-2 39.063 31.467 57.945 49.905 114.000 106.118 97.664 89.778
R65-3 36.892 23.872 57.323 42.383 112.024 88.953 95.912 75.141
R65-4 33.637 21.701 53.069 36.884 81.534 54.973 72.402 48.929
R65-5 34.722 20.616 48.348 32.422 57.901 46.979 54.232 42.150
R65-6 34.722 18.446 46.481 28.843 39.697 30.790 40.481 29.394
R86-1 42.318 32.552 257.408 182.032 147.280 99.575 157.938 108.538
R86-2 40.148 27.127 134.307 110.392 69.436 63.268 78.480 68.574
R86-3 39.063 21.701 116.565 109.562 66.584 63.193 73.131 67.910
R86-4 37.977 20.616 109.925 105.930 59.027 54.548 66.287 60.799
R86-5 36.892 19.531 106.190 100.950 58.577 53.322 65.199 58.918
R86-6 32.552 17.361 55.870 54.625 46.491 43.814 46.974 43.487
R115-1 22.690 19.353 41.415 39.756 52.774 46.321 48.224 42.815
R115-2 20.688 16.684 40.296 37.556 44.978 38.630 42.015 36.452
R115-3 20.021 14.014 39.640 36.861 39.192 37.052 37.509 34.914
R115-4 20.688 12.680 33.773 20.804 31.550 25.203 30.908 23.361
R115-5 18.686 10.678 24.857 18.681 25.748 25.162 24.961 22.812
R135-1 26.027 20.021 71.753 56.584 63.760 42.341 61.572 42.552
R135-2 24.025 16.016 58.437 49.675 43.871 33.707 44.359 34.614
R135-3 18.019 12.012 28.987 19.839 38.622 26.358 35.213 24.015
R135-4 17.351 10.010 36.938 18.180 34.789 25.610 33.534 23.007
R135-5 16.016 8.676 31.843 16.751 25.227 24.348 25.431 21.713
299
Table 3 Contd:
Identification TPO43-
(mg/L/g) DPO4
3- (mg/L/g)
TP (mg/L/g)
DTP (mg/L/g)
R20-1 1468.565 925.123 908.703 351.646
R20-2 1425.076 521.595 859.230 343.195
R20-3 1399.553 464.399 866.165 297.462
R20-4 1234.700 437.502 694.133 71.651
R20-5 488.411 63.173 510.556 57.155
R20-6 314.180 37.168 267.257 94.295
R40-1 841.339 127.032 598.544 257.896
R40-2 775.590 108.414 521.130 218.068
R40-3 386.284 73.925 231.492 95.346
R40-4 380.787 72.241 230.259 70.782
R40-5 197.934 52.627 126.825 47.431
R40-6 125.383 43.798 105.415 32.126
R65-1 13.319 6.354 13.438 9.996
R65-2 0.865 0.865 8.859 4.259
R65-3 0.865 0.865 6.095 3.031
R65-4 0.865 0.865 5.214 2.444
R65-5 0.865 0.865 5.022 1.863
R65-6 0.865 0.865 4.648 0.950
R86-1 14.414 4.884 19.853 5.791
R86-2 4.497 2.162 4.777 2.268
R86-3 2.479 0.836 3.939 1.163
R86-4 0.865 0.865 3.122 1.115
R86-5 0.865 0.865 2.007 0.560
R86-6 0.865 0.865 0.619 0.448
R115-1 152.750 30.896 180.854 94.228
R115-2 140.645 30.767 196.161 89.187
R115-3 99.143 30.711 129.334 50.024
R115-4 53.491 20.479 99.260 47.236
R115-5 45.398 16.804 89.363 4.233
R135-1 212.052 45.472 227.748 62.247
R135-2 163.347 43.803 133.533 61.225
R135-3 50.642 27.751 109.693 51.546
R135-4 40.418 11.182 45.543 15.661
R135-5 34.239 7.581 44.149 11.578
Note: R- roof surfaces; TTU- Turbidity; EC- Electrical conductivity; TNO2-- Total nitrite-nitrogen;
DNO2-- Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon;
TNO3-- Total nitrate- nitrogen; DNO3
-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen;
DKN- Dissolved kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total
suspended solids; TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3--
Dissolved Total Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
300
301
Appendix E
Data Matrices for Validation
302
303
Table 1- Data set 1-Armstrong Drive (Miguntanna N. 2009- unpublished data)
Sample name Intensity(mm/hr) and
duration (min) Identification
EC
(µS/cm) TSS
(mg/L) TDS
(mg/L) TS
(mg/L) TOC
(mg/L) DOC
(mg/L)
A-20-0-5 A20-1 44.0 28.42 135.00 163.42 17.330 13.760
A-20-5-10 A20-2 40.2 26.66 132.50 159.16 10.290 8.270
A-20-10-15 A20-3 39.1 26.00 122.50 148.50 9.538 6.361
A-20-15-20 A20-4 36.1 14.00 90.00 104.00 8.351 4.935
A-20-20-25 A20-5 33.6 21.00 117.50 138.50 7.494 3.926
A-20-25-30 A20-6 32.9 21.00 115.00 136.00 6.866 3.975
A-20-30-35 A20-7 31.5 18.00 102.50 120.50 6.414 3.594
A-20-35-40 A20-8 28.2 13.00 100.00 113.00 5.926 3.390
A-40-0-5 A40-1 42.8 30.00 130.00 160.00 18.740 11.650
A-40-5-10 A40-2 36.7 29.00 112.50 141.50 14.685 11.610
A-40-10-15 A40-3 37.0 27.00 110.00 137.00 13.667 10.190
A-40-15-20 A40-4 36.2 23.00 110.00 133.00 12.903 9.664
A-40-20-25 A40-5 37.3 21.00 107.50 128.50 10.169 8.156
A-40-25-30 A40-6 35.7 10.00 57.50 67.50 10.689 8.549
A-40-30-35 A40-7 34.5 8.00 50.00 58.00 10.209 6.865
A-65-0-5 A65-1 36.8 51.80 150.00 201.80 14.653 10.130
A-65-5-10 A65-2 18.6 42.92 112.50 155.42 10.947 5.983
A-65-10-15 A65-3 20.0 39.96 82.50 122.46 9.846 5.799
A-65-15-20 A65-4 22.9 34.04 75.00 109.04 9.738 6.213
A-65-20-25 A65-5 22.9 31.08 67.50 98.58 9.121 6.039
A-65-25-30 A65-6 21.3 14.80 22.50 37.30 8.738 5.167
A-86-0-5 A86-1 45.9 58.90 187.50 246.40 23.840 15.520
A-86-5-10 A86-2 30.0 29.70 110.00 139.70 19.175 12.900
A-86-10-15 A86-3 30.9 21.60 102.50 124.10 16.693 10.600
A-86-15-20 A86-4 30.5 21.06 92.50 113.56 15.073 8.781
A-115-0-5 A115-1 34.4 69.44 175.25 244.69 24.490 12.870
A-115-5-10 A115-2 31.4 38.44 132.50 170.94 11.265 7.530
A-115-10-15 A115-3 31.4 13.64 112.50 126.14 8.085 5.130
A-135-0-5 A115-4 35.6 73.44 222.50 295.94 12.550 8.151
A-135-5-10 A135-1 35.5 13.50 97.50 111.00 7.360 4.111
A-135-10-15 A135-2 37.2 17.28 92.50 109.78 6.478 3.695
A-135-15-20 A135-3 35.9 16.20 92.50 108.70 6.081 3.339
304
Table 1 Contd:
Identification
TNO2--
(mg/L) DNO2
- (mg/L)
TNO3-
(mg/L) DNO3
- (mg/L)
TKN (mg/L)
DKN (mg/L)
TN (mg/L)
DTN (mg/L)
A20-1 0.005 0.004 0.150 0.137 3.581 3.067 3.736 3.208
A20-2 0.003 0.003 0.092 0.088 2.785 1.698 2.880 1.789
A20-3 0.003 0.002 0.091 0.084 1.365 1.324 1.459 1.410
A20-4 0.002 0.001 0.044 0.053 1.016 0.950 1.062 1.004
A20-5 <0.001 0.000 0.055 0.049 0.757 0.520 0.812 0.569
A20-6 <0.001 0.000 0.046 0.020 0.649 0.536 0.695 0.556
A20-7 <0.001 0.000 0.033 0.023 0.581 0.349 0.614 0.372
A20-8 <0.001 0.000 0.018 0.011 0.399 0.333 0.417 0.344
A40-1 0.007 0.005 0.188 0.187 3.486 2.933 3.681 3.125
A40-2 0.004 0.003 0.165 0.118 2.593 2.251 2.762 2.372
A40-3 0.003 0.002 0.151 0.104 1.635 1.687 1.789 1.793
A40-4 0.001 0.001 0.093 0.081 1.558 1.554 1.652 1.636
A40-5 0.001 0.001 0.062 0.049 1.458 1.373 1.521 1.423
A40-6 <0.001 0.000 0.029 0.029 1.386 1.207 1.415 1.236
A40-7 <0.001 0.000 0.028 0.019 1.281 0.953 1.309 0.972
A65-1 0.006 0.004 0.289 0.207 2.042 1.559 2.337 1.770
A65-2 0.002 0.001 0.223 0.222 1.094 1.092 1.319 1.315
A65-3 <0.001 0.000 0.220 0.186 0.800 0.556 1.020 0.742
A65-4 <0.001 0.000 0.210 0.199 0.718 0.635 0.928 0.834
A65-5 <0.001 0.000 0.198 0.193 0.636 0.621 0.834 0.814
A65-6 <0.001 0.000 0.197 0.197 0.597 0.534 0.794 0.731
A86-1 0.008 0.007 0.096 0.106 2.880 3.724 2.984 3.837
A86-2 0.007 0.006 0.084 0.076 2.792 2.752 2.883 2.834
A86-3 0.005 0.001 0.106 0.106 2.040 1.941 2.151 2.048
A86-4 <0.001 0.000 0.005 0.000 1.519 1.352 1.524 1.352
A115-1 0.006 0.004 0.051 0.049 2.847 2.539 2.904 2.592
A115-2 0.005 0.003 0.032 0.025 1.497 1.249 1.534 1.277
A115-3 0.002 0.001 0.029 0.025 0.693 0.656 0.724 0.682
A115-4 0.007 0.006 0.073 0.063 2.212 1.677 2.292 1.746
A135-1 0.005 0.003 0.035 0.035 0.951 0.933 0.991 0.971
A135-2 0.001 0.001 0.022 0.029 0.624 0.615 0.647 0.645
A135-3 <0.001 0.000 0.015 0.013 0.603 0.559 0.618 0.572
305
Table 1 Contd:
Identification
TPO43-
(mg/L) DPO4
3- (mg/L)
TP (mg/L)
DTP (mg/L)
A20-1 0.135 0.014 0.157 0.014
A20-2 0.022 0.009 0.136 0.010
A20-3 0.017 0.008 0.134 0.068
A20-4 0.015 0.007 0.076 0.008
A20-5 0.012 0.003 0.073 0.005
A20-6 0.010 0.003 0.068 0.005
A20-7 0.008 0.003 0.059 0.005
A20-8 0.007 0.003 0.040 0.005
A40-1 0.151 0.009 0.567 0.010
A40-2 0.057 0.008 0.249 0.009
A40-3 0.044 0.007 0.245 0.008
A40-4 0.004 0.005 0.105 0.005
A40-5 0.003 0.003 0.029 0.005
A40-6 0.002 0.003 0.143 0.005
A40-7 0.003 0.007 0.120 0.008
A65-1 0.126 0.006 0.679 0.020
A65-2 0.035 0.005 0.661 0.012
A65-3 0.025 0.003 0.631 0.005
A65-4 0.013 0.003 0.613 0.005
A65-5 0.012 0.003 0.512 0.005
A65-6 0.017 0.003 0.514 0.005
A86-1 0.129 0.013 0.862 0.014
A86-2 0.076 0.012 0.651 0.014
A86-3 0.065 0.005 0.645 0.005
A86-4 0.030 0.005 0.612 0.005
A115-1 0.096 0.026 0.911 0.027
A115-2 0.085 0.014 0.887 0.016
A115-3 0.007 0.005 0.070 0.005
A115-4 0.042 0.039 0.928 0.049
A135-1 0.022 0.029 0.923 0.039
A135-2 0.017 0.025 0.515 0.027
A135-3 0.091 0.012 0.291 0.014 Note: A- Amstrong Drive; EC- Electrical conductivity; TNO2
-- Total nitrite-nitrogen; DNO2--
Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3-- Total
nitrate- nitrogen; DNO3-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN- Dissolved
kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids;
TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3-- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
306
Table 2- Data set 2-Stevens Street (Miguntanna N. 2009- unpublished data)
Sample name Intensity(mm/hr) and
duration (min) Identification
EC
(µS/cm) TSS
(mg/L) TDS
(mg/L) TS
(mg/L) TOC
(mg/L) DOC
(mg/L)
S-20-0-5 S20-1 37.5 498.80 340.00 838.80 11.750 10.380
S-20-5-10 S20-2 30.4 463.40 292.00 755.40 9.911 6.400
S-20-10-15 S20-3 23.9 240.53 272.00 512.53 8.872 4.826
S-20-15-20 S20-4 22.9 157.00 268.00 425.00 6.058 4.237
S-20-20-25 S20-5 23.7 125.36 210.00 335.36 5.307 3.550
S-20-25-30 S20-6 21.7 120.40 198.00 318.40 4.981 3.736
S-20-30-35 S20-7 20.2 101.14 182.00 283.14 4.881 3.890
S-40-0-5 S40-1 161.8 621.60 376.00 997.60 25.870 18.670
S-40-5-10 S40-2 37.6 615.60 338.00 953.60 12.355 8.513
S-40-10-20 S40-3 30.6 256.40 316.00 572.40 8.864 5.395
S-40-20-25 S40-4 27.4 251.76 272.00 523.76 7.410 4.817
S-40-25-30 S40-5 28.3 242.07 266.00 508.07 7.141 4.913
S-65-0-5 S65-1 132.6 692.00 378.00 1070.00 28.800 11.330
S-65-5-10 S65-2 82.1 359.80 258.00 617.80 7.333 3.900
S-65-10-15 S65-3 21.6 355.20 150.00 505.20 5.874 2.613
S-65-15-20 S65-4 16.65 312.20 140.00 452.20 4.806 2.348
S-65-20-25 S65-5 20.8 179.52 122.00 301.52 4.111 2.175
S-65-25-30 S65-6 15.2 105.67 50.00 155.67 3.743 2.038
S-86-0-5 S86-1 65.8 778.80 388.00 1166.80 20.320 15.550
S-86-5-10 S86-2 21.1 338.20 176.00 514.20 8.777 5.994
S-86-10-15 S86-3 12.86 265.73 128.00 393.73 6.635 4.366
S-86-15-20 S86-4 19.78 227.90 116.00 343.90 10.230 8.788
S-86-20-25 S86-5 9.8 127.84 102.00 229.84 8.492 5.907
S-115-0-5 S115-1 35.8 797.60 438.00 1235.60 19.620 12.640
S-115-5-10 S115-2 8.8 368.00 362.00 730.00 6.022 4.211
S-115-10-15 S115-3 7.88 257.87 306.00 563.87 4.588 3.164
S-115-15-20 S115-4 5.4 129.30 198.00 327.30 2.690 2.658
S-135-0-5 S135-1 39.4 816.40 584.00 1400.40 15.620 7.502
S-135-5-10 S135-2 21.11 404.80 270.00 674.80 5.945 2.764
S-135-10-15 S135-3 18.23 286.67 220.00 506.67 6.616 3.415
S-135-15-20 S135-4 16.6 157.00 210.00 367.00 4.230 2.980
307
Table 2 Contd:
Identification
TNO2-
(mg/L) DNO2
- (mg/L)
TNO3-
(mg/L) DNO3
- (mg/L)
TKN (mg/L)
DKN (mg/L)
TN (mg/L)
DTN (mg/L)
S20-1 0.004 0.002 0.284 0.161 2.612 1.507 2.900 1.670
S20-2 0.003 0.001 0.159 0.154 1.395 1.138 1.557 1.293
S20-3 0.001 0.000 0.097 0.062 1.120 1.103 1.218 1.165
S20-4 <0.001 0.000 0.074 0.059 0.797 0.724 0.872 0.783
S20-5 <0.001 0.000 0.059 0.051 0.700 0.593 0.760 0.644
S20-6 <0.001 0.000 0.046 0.048 0.613 0.548 0.660 0.596
S20-7 <0.001 0.000 0.035 0.039 0.587 0.411 0.622 0.450
S40-1 0.011 0.000 0.274 0.283 2.587 2.255 2.872 2.538
S40-2 0.001 0.001 0.153 0.191 1.609 1.624 1.763 1.816
S40-3 0.000 0.000 0.121 0.087 1.564 1.056 1.684 1.143
S40-4 0.001 0.000 0.104 0.113 0.984 0.983 1.089 1.096
S40-5 <0.001 0.002 0.101 0.090 0.852 0.841 0.953 0.931
S65-1 0.003 0.002 0.272 0.137 2.000 1.084 2.275 1.223
S65-2 0.002 0.001 0.258 0.128 0.790 0.771 1.050 0.900
S65-3 <0.001 0.000 0.220 0.092 0.756 0.608 0.976 0.700
S65-4 0.001 0.001 0.200 0.078 0.729 0.629 0.930 0.708
S65-5 0.001 0.001 0.179 0.076 0.974 0.569 1.154 0.646
S65-6 0.001 0.001 0.163 0.052 0.615 0.605 0.780 0.658
S86-1 0.006 0.001 0.383 0.222 2.365 1.579 2.754 1.802
S86-2 0.001 0.003 0.278 0.116 1.440 1.265 1.719 1.384
S86-3 0.001 0.001 0.239 0.092 1.339 0.956 1.579 1.049
S86-4 <0.001 0.001 0.218 0.088 2.371 1.861 2.588 1.949
S86-5 <0.001 0.000 0.196 0.069 1.438 1.424 1.634 1.493
S115-1 0.010 0.000 0.392 0.272 1.576 1.707 1.978 1.979
S115-2 0.001 0.006 0.164 0.096 0.980 0.932 1.145 1.034
S115-3 <0.001 0.001 0.106 0.058 0.744 0.722 0.850 0.780
S115-4 <0.001 0.000 0.074 0.046 0.571 0.513 0.644 0.559
S135-1 0.013 0.008 0.268 0.261 1.427 0.965 1.708 1.234
S135-2 0.001 0.001 0.105 0.094 0.656 0.654 0.762 0.749
S135-3 <0.001 0.000 0.056 0.054 0.530 0.497 0.586 0.551
S135-4 <0.001 0.005 0.042 0.048 0.334 0.310 0.376 0.358
308
Table 2 Contd:
Identification
TPO43-
(mg/L) DPO4
3- (mg/L)
TP (mg/L)
DTP (mg/L)
S20-1 1.385 0.499 4.857 0.520
S20-2 0.943 0.430 4.391 0.434
S20-3 0.733 0.347 4.246 0.424
S20-4 0.783 0.328 1.399 0.412
S20-5 0.684 0.307 1.387 0.384
S20-6 0.632 0.307 1.285 0.292
S20-7 0.621 0.302 1.279 0.270
S40-1 1.537 0.531 5.993 0.823
S40-2 1.720 0.432 4.380 0.502
S40-3 1.406 0.381 4.231 0.449
S40-4 1.242 0.360 4.243 0.442
S40-5 1.202 0.346 2.209 0.326
S65-1 1.616 0.595 7.711 0.830
S65-2 1.234 0.482 6.439 0.488
S65-3 0.879 0.384 4.287 0.387
S65-4 0.733 0.334 3.216 0.385
S65-5 0.155 0.031 0.816 0.170
S65-6 <0.03 <0.005 0.756 0.155
S86-1 1.648 0.688 8.904 0.843
S86-2 1.343 0.372 4.962 0.527
S86-3 1.201 0.321 4.609 0.372
S86-4 0.840 0.317 4.264 0.331
S86-5 0.853 0.309 1.190 0.315
S115-1 2.065 0.691 9.465 0.855
S115-2 1.277 0.328 5.127 0.673
S115-3 1.093 0.296 4.159 0.344
S115-4 0.651 0.281 1.842 0.258
S135-1 2.706 0.736 9.870 1.282
S135-2 1.038 0.386 4.569 0.447
S135-3 0.865 0.388 4.168 0.427
S135-4 0.562 0.368 4.136 0.386 Note: S- Stevens Street; EC- Electrical conductivity; TNO2
-- Total nitrite-nitrogen; DNO2-- Dissolved
nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3-- Total nitrate-
nitrogen; DNO3-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN- Dissolved kjeldahl
nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids; TDS-
Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3-- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
309
Table 3- Data set 3- Lawrence Drive (Miguntanna N. 2009- unpublished data)
Sample name Intensity(mm/hr) and duration (min)
Identification
EC (µS/cm)
TSS (mg/L)
TDS (mg/L)
TS (mg/L)
TOC (mg/L)
DOC (mg/L)
L-20-0-5 L20-1 502.0 295.60 792.50 1088.10 67.840 61.440
L-20-5-10 L20-2 431.0 207.00 750.00 957.00 29.680 26.200
L-20-10-15 L20-3 438.0 192.00 730.00 922.00 23.850 19.810
L-20-15-20 L20-4 428.0 135.20 797.50 932.70 20.340 16.620
L-20-20-25 L20-5 413.0 134.40 782.50 916.90 18.140 14.940
L-20-25-30 L20-6 419.0 99.20 705.00 804.20 18.050 15.370
L-20-30-35 L20-7 408.0 91.20 447.50 538.70 17.310 13.740
L-20-35-40 L20-8 401.0 90.80 177.50 268.30 13.200 10.930
L-40-0-5 L40-1 482.0 307.60 925.00 1232.60 57.020 49.050
L-40-5-10 L40-2 490.0 265.60 842.50 1108.10 33.860 28.800
L-40-10-15 L40-3 451.0 178.80 802.50 981.30 25.840 21.480
L-40-15-20 L40-4 487.0 160.80 822.50 983.30 19.570 16.730
L-40-20-25 L40-5 443.0 82.80 775.00 857.80 16.890 13.910
L-40-25-30 L40-6 448.0 60.80 742.50 803.30 14.900 11.700
L-40-30-35 L40-7 394.0 62.00 717.50 779.50 13.880 10.710
L-65-0-5 L65-1 612.0 310.60 940.00 1250.60 27.100 18.180
L-65-5-10 L65-2 471.0 250.40 227.50 477.90 16.380 11.180
L-65-10-15 L65-3 472.0 129.20 217.50 346.70 13.150 9.538
L-65-15-20 L65-4 460.0 77.20 207.50 284.70 10.255 8.015
L-65-20-25 L65-5 486.0 71.20 205.00 276.20 9.456 7.236
L-65-25-30 L65-6 461.0 49.40 182.50 231.90 8.474 8.648
L-86-0-5 L86-1 498.0 335.60 872.50 1208.10 31.280 26.280
L-86-5-10 L86-2 474.0 128.80 832.50 961.30 15.170 11.720
L-86-10-15 L86-3 484.0 68.40 837.50 905.90 15.121 9.662
L-86-15-20 L86-4 474.0 57.20 757.50 814.70 10.810 8.919
L-86-20-25 L86-5 544.0 56.40 680.00 736.40 10.779 8.062
L-115-0-5 L115-1 448.0 374.00 895.00 1269.00 40.200 35.010
L-115-5-10 L115-2 493.0 220.40 872.50 1092.90 18.410 13.940
L-115-10-15 L115-3 452.0 242.00 860.00 1102.00 14.150 10.700
L-115-15-20 L115-4 482.0 146.00 850.00 996.00 10.002 8.479
L-135-0-5 L135-1 387.0 428.00 987.50 1415.50 14.660 10.930
L-135-5-10 L135-2 354.0 229.60 952.50 1182.10 8.585 5.689
L-135-10-15 L135-3 332.0 61.60 772.50 834.10 7.468 5.233
L-135-15-20 L135-4 319.0 36.80 715.00 751.80 6.217 4.896
310
Table 3 Contd:
Identification
TNO2-
(mg/L) DNO2
- (mg/L)
TNO3-
(mg/L) DNO3
- (mg/L)
TKN (mg/L)
DKN (mg/L)
TN (mg/L)
DTN (mg/L)
L20-1 0.138 0.122 0.905 0.498 15.532 11.196 16.575 11.815
L20-2 0.038 0.037 0.433 0.242 6.737 4.993 7.208 5.273
L20-3 0.035 0.031 0.396 0.198 5.095 4.124 5.526 4.353
L20-4 0.025 0.024 0.359 0.184 4.935 3.452 5.319 3.660
L20-5 0.026 0.025 0.355 0.182 3.408 3.245 3.789 3.452
L20-6 0.023 0.023 0.248 0.134 3.374 2.960 3.645 3.117
L20-7 0.028 0.026 0.241 0.150 3.300 3.113 3.569 3.289
L20-8 0.019 0.019 0.208 0.126 2.423 1.996 2.650 2.141
L40-1 0.075 0.058 0.903 0.567 12.499 9.948 13.477 10.573
L40-2 0.027 0.027 0.381 0.171 6.492 5.206 6.900 5.404
L40-3 0.025 0.024 0.376 0.144 4.999 3.980 5.400 4.148
L40-4 0.023 0.020 0.264 0.136 3.800 3.289 4.087 3.445
L40-5 0.018 0.018 0.232 0.231 2.877 2.664 3.126 2.913
L40-6 0.016 0.015 0.234 0.126 2.479 2.353 2.729 2.494
L40-7 0.013 0.011 0.208 0.126 1.932 1.960 2.154 2.097
L65-1 0.046 0.045 0.653 0.301 5.885 3.624 6.584 3.970
L65-2 0.026 0.025 0.417 0.197 2.875 2.158 3.317 2.380
L65-3 0.020 0.019 0.408 0.192 1.830 1.640 2.258 1.850
L65-4 0.018 0.018 0.397 0.190 1.747 1.102 2.162 1.310
L65-5 0.015 0.014 0.382 0.171 1.565 0.973 1.962 1.159
L65-6 0.014 0.012 0.202 0.145 0.886 0.732 1.102 0.889
L86-1 0.038 0.035 0.738 0.396 5.920 4.963 6.696 5.394
L86-2 0.018 0.018 0.656 0.382 2.275 1.965 2.948 2.365
L86-3 0.018 0.017 0.654 0.372 1.272 1.216 1.944 1.604
L86-4 0.017 0.016 0.277 0.168 1.155 1.051 1.449 1.235
L86-5 0.016 0.016 0.253 0.155 0.958 0.862 1.228 1.033
L115-1 0.039 0.036 0.280 0.168 3.180 6.225 3.499 6.430
L115-2 0.025 0.021 0.236 0.138 3.103 2.802 3.364 2.961
L115-3 0.022 0.020 0.229 0.161 2.088 1.983 2.339 2.163
L115-4 0.018 0.017 0.213 0.119 1.298 1.245 1.528 1.381
L135-1 0.021 0.018 0.285 0.199 2.066 1.993 2.372 2.210
L135-2 0.015 0.012 0.240 0.125 1.101 1.097 1.356 1.234
L135-3 0.012 0.010 0.232 0.121 0.521 0.563 0.764 0.694
L135-4 0.009 0.009 0.218 0.118 0.480 0.703 0.707 0.830
311
Table 3 Contd:
Identification
TPO43-
(mg/L) DPO4
3- (mg/L)
TP (mg/L)
DTP (mg/L)
L20-1 1.289 0.281 4.453 0.389
L20-2 1.158 0.275 4.455 0.334
L20-3 1.134 0.268 4.450 0.275
L20-4 1.058 0.267 4.449 0.274
L20-5 0.771 0.256 3.473 0.265
L20-6 0.694 0.259 3.127 0.263
L20-7 0.672 0.256 2.616 0.259
L20-8 <0.03 <0.03 0.724 0.162
L40-1 0.901 0.258 4.462 0.270
L40-2 0.872 0.257 3.260 0.263
L40-3 0.635 0.256 3.232 0.259
L40-4 0.618 0.256 2.594 0.268
L40-5 0.549 0.255 1.757 0.266
L40-6 0.521 0.255 1.739 0.262
L40-7 0.512 0.254 1.701 0.254
L65-1 0.942 0.405 4.546 0.412
L65-2 0.852 0.395 4.159 0.403
L65-3 0.785 0.268 4.802 0.346
L65-4 0.719 0.267 4.480 0.281
L65-5 0.568 0.263 3.062 0.318
L65-6 <0.03 <0.005 0.371 0.318
L86-1 0.979 0.266 5.314 0.379
L86-2 0.858 0.262 4.325 0.366
L86-3 0.645 0.265 4.278 0.340
L86-4 0.536 0.261 4.160 0.218
L86-5 <0.03 <0.005 1.556 0.166
L115-1 1.408 0.370 5.808 0.374
L115-2 1.347 0.260 4.853 0.362
L115-3 1.252 0.254 3.549 0.259
L115-4 <0.03 <0.005 2.642 0.154
L135-1 2.621 1.492 10.393 1.512
L135-2 1.254 0.258 9.129 0.878
L135-3 0.987 0.285 5.120 0.425
L135-4 0.856 0.251 4.131 0.364 Note: L- Lawrence Drive; EC- Electrical conductivity; TNO2
-- Total nitrite-nitrogen; DNO2--
Dissolved nitrite-nitrogen TOC- Total organic carbon; DOC- Dissolved organic carbon; TNO3-- Total
nitrate- nitrogen; DNO3-- Dissolved nitrate- nitrogen; TKN- Total kjeldahl nitrogen; DKN- Dissolved
kjeldahl nitrogen; TN- Total nitrogen; DTN- Dissolved total nitrogen; TSS- Total suspended solids;
TDS- Total dissolved solids; TS- Total solids; TPO43-- Total Phosphates; DPO4
3-- Dissolved Total
Phosphates; TP- Total phosphorus; DTP- Dissolved total phosphorus.
