modelling of air quality in street canyons · 2017-10-27 · modelling of air quality in street...

MODELLING OF AIR QUALITY IN STREET CANYONS

by

IOSIF A. KAROUSOS, M.Sc.(Eng)

This Dissertation is submitted by the Candidate

in Partial Fulfilment of the Requirements for the Degree of

MSc(Eng) Transport Planning and Engineering

Submission by the Candidate does not imply

that its content or standard is endorsed by

the Examiners

Institute for Transport Studies

The University of Leeds

August 2006

ii

Abstract

Despite the recent technological improvements on vehicle emission control and the

increasingly stricter regulations, traffic-related air pollution is one of the most pressing

problems in modern urban agglomerations both in the developed and the developing

world. Air quality limit values, whose objective is to protect public health, are

frequently exceeded, particularly in busy streets and urban areas. Evidence is constantly

emerging related to the human exposure to increased pollutant concentrations in

densely populated urban areas in contrast with the proven adverse effects on human

health. It is, hence, imperative to fully understand the pollutants behaviour within

confined urban surroundings in order to achieve further improvements in urban air

quality.

This study presents the results of a limited monitoring and modelling methodology,

which is adopted in order to understand the predominant mechanisms of pollution

dispersion in an urban street canyon. The examined canyon in Headingley area of Leeds

is relatively narrow and flanked by closely spaced buildings on both sides. Evidence of

the street canyon effect on the diffusion of Carbon Monoxide emissions, hence, has been

attempted to be established via the analysis of the measured levels and the results of the

application of OSPM dispersion model. The relationship between the engine emission

rates, the ambient conditions and other factors affecting the pollutants dispersion could

not be spherically examined because of the shortcomings in both the monitoring and the

modelling procedure. Moreover, the model predictive performance of CO levels has

been proven fairly poor, when evaluated against the observed concentrations.

WinOSPM generally underpredicted the CO measurements and its sensitivity to wind

speed and direction changes could not reveal lucid evidence of the street canyon effect.

iii

Table of Contents

Abstract......................................................................................................................................... ii

Table of Contents ...................................................................................................................... iii

List of Tables ...............................................................................................................................v

List of Figures .............................................................................................................................vi

Chapter 1 Introduction ............................................................................................................ 1

1.1 Background ................................................................................................................... 1

1.2 Objectives and Scope of the Study.............................................................................. 2

1.3 Structure of the Dissertation ....................................................................................... 4

Chapter 2 Traffic Pollution ..................................................................................................... 6

2.1 Main types of air pollutants......................................................................................... 7

2.2 National Air Quality Strategy ................................................................................... 13

2.3 Vehicle Emissions ....................................................................................................... 16

2.4 Dispersion of air pollution in urban areas............................................................... 20

2.5 Monitoring traffic pollution ...................................................................................... 24

2.6 Modelling traffic pollution......................................................................................... 26

2.6.1 Types of air quality models ........................................................................... 26

Chapter 3 Study Methodology ............................................................................................. 30

3.1 Monitoring methodology........................................................................................... 30

3.1.1 Site and equipment description ................................................................... 31

3.1.2 Data collection.................................................................................................. 34

3.1.2.1 Concentrations Data ................................................................................... 35

3.1.2.2 Meteorology and Background Data ............................................................. 36

3.1.2.3 Traffic Flows ................................................................................................ 36

3.2 Modelling procedure .................................................................................................. 38

3.2.1 Parameterisation of`WinOSPM..................................................................... 38

3.3 Data analysis ............................................................................................................... 43

iv

Chapter 4 Monitoring and Modelling Results ................................................................. 44

4.1 Traffic and meteorology ............................................................................................. 44

4.1.1 Traffic Flow ...................................................................................................... 44

4.1.2 Vehicle Type Distribution .............................................................................. 47

4.1.3 Meteorological Data........................................................................................ 48

4.2 Streetboxes data analysis ........................................................................................... 49

4.3 Monitored versus modelled results ......................................................................... 53

4.4 Summary ..................................................................................................................... 59

Chapter 5 Discussion and Recommendations .................................................................. 61

5.1 Synopsis ........................................................................................................................ 61

5.2 Conclusions and discussion ...................................................................................... 62

5.3 Recommendations for further work ........................................................................ 65

Acknowledgements ................................................................................................................. 67

References .................................................................................................................................. 68

Appendices ................................................................................................................................ 71

APPENDIX I: MANUAL TRAFFIC SURVEY FORM .................................................. 71

APPENDIX II: STATISTICAL ANALYSIS RESULTS .................................................. 72

APPENDIX III: fluidyn-PANACHE SETUP ................................................................. 75

v

List of Tables

Table 2.1 Air pollutant types (including greenhouse gases). 8 Table 2.2 New air quality objectives included in the NAQS for protecting human

health. 15

Table 2.3 Environmental policy issues and analogous scales of dispersion phenomena. 21

Table 2.4 Main types of existing air quality models. 27

Table 3.1 LEARIAN Streetbox characteristics. 34

Table 4.1 Original, factored-up and completed hourly traffic flows. 46

Table 4.2 WinOSPM Results Summary Table. 53

Table 4.3 Results of statistical analyses between observed and predicted CO values. 57

Table 4.4 Wind speed dependence of the observed and predicted average CO levels. 58

vi

List of Figures

Figure 1.1 The location of the examined street canyon in Headingley, Leeds. 3

Figure 2.1 Estimated CO emissions by source. UK, 1970-2004. 9

Figure 2.2 Estimated CO emissions by vehicle type. UK, 1970-2004. 9

Figure 2.3 Estimated NOx emissions by vehicle type. UK, 1970-2004. 10

Figure 2.4 Estimated PM10 Emissions by vehicle type. UK, 1970-2004. 12

Figure 2.5 Total and traffic-related emissions of key air pollutants in UK for 2004. 17

Figure 2.6 Flow chart of the application of COPERT-III baseline methodology. 20

Figure 2.7 Perpendicular flow systems in urban areas for different road widths. 22 Figure 2.8 Vertical cross-section of a typical symmetric urban street canyon. The

recirculating wind flow is shown in the case of perpendicular roof level wind. 23

Figure 3.1 View of the street canyon from the Otley Road/North Lane intersection. 31

Figure 3.2 Map of the immediate neighbourhood of the examined street canyon and of the utilised monitoring points. 32

Figure 3.3 Vertical cross-section of the street canyon showing the monitoring points on the lampposts. 33

Figure 3.4 LEARIAN Streetbox No 133. Enclosure and example of fixing on a lamppost. 34

Figure 3.5 Looking north towards the junction and the locations of three Streetboxes. 35

Figure 3.6 The utilised tally counter & the view from a survey point. 37

Figure 3.7 Topographical features of the Otley Road street canyon. 39 Figure 3.8 Traffic Data Editor window for the hourly traffic distribution in Otley

Road. 40

vii

Figure 3.9 (a) Average diurnal traffic flows & (b) CO emissions for three vehicle categories (adapted from WinOSPM Traffic Editor window). 41

Figure 3.10 WinOSPM configuration window for Vehicle Emission Factors. 42

Figure 3.11 Variables list in the weather and background input file of WinOSPM. 42 Figure 4.1 Observed fractional traffic counts and average diurnal profile on Otley

Road. 45 Figure 4.2 Completed diurnal flows and associated typical profile for AADT ≈

35,884. 46 Figure 4.3 Estimated vehicle distribution in Headingley for (a) 2006 and (b) 2004,

and (c) for current UK national urban conditions. 47

Figure 4.4 The frequency distribution of the wind speed measurements. 49

Figure 4.5 Wind rose for the examined time period. 49

Figure 4.6 Average measured CO concentration by the installed Streetboxes. 51 Figure 4.7 Average diurnal CO measurements at each Streetbox vs. traffic flow

profile. 51

Figure 4.8 Mean CO concentrations for the examined wind sectors. 52 Figure 4.9 Average diurnal CO predictions at each receptor point vs. traffic flow

profile. 54 Figure 4.10 Observed and predicted hourly CO concentrations from Monday, 15th

until Sunday, 21st May 2006. 55 Figure 4.11 Scatter plots of measured against predicted CO values for the two

pairs of monitoring points. 57 Figure 4.12 The dependence of the average CO predictions on combined wind

speed and direction clusters. The shaded bars indicate the leeward side in each case. 59

- 1 -

Chapter 1

Introduction

1.1 Background

The good condition of ambient air is a significant factor affecting quality of everyday

life, particularly in large conurbations. The population living in modern cities is

constantly increasing worldwide and, on the other hand, the quality of urban

environment is deteriorating. Air pollution is not a new problem, but the types and

behaviour of the various air pollutants have changed during the last decades. Severe

environmental problems have been present since the Industrial Revolution, mostly

due to the intense use of fossil fuels in the industry and domestic sector. However,

since the UK Government’s Clean Air Act implemented in 1956, these emissions have

been significantly reduced and over the last two decades the main source of air

pollution has been attributed to the motorised traffic. Increasing vehicle volumes,

heavier congestion and more complex urban topographies outline the key causes of

this alteration (Michail, 2003).

Nowadays, national and local authorities are trying to implement policy instruments

and traffic management schemes to tackle poor air quality issues and locate the

contamination ‘hot spots’, mainly in developed countries and large cities. Parallel to

stricter legislations and reduced vehicle emissions, systematic air quality monitoring

and modelling are essential tools in order to understand the behaviour of various

toxic substances emitted by motor vehicles.

Monitoring campaigns can, however, be costly and time-consuming procedures if

extensive and accurate results are to be achieved. That is why over the last years

I.A. Karousos Modelling of Air Quality in Street Canyons 1. Introduction

- 2 -

various types of air quality models have been developed, which are able to assess

roadside air quality by predicting current and future pollutant levels and providing

temporal and spatial variations for a wide range of topographies and conditions

(Sharma and Khare, 2001). These mathematical models are able to study the physical

and chemical processes related to the dispersion and transformation of gaseous

pollutants and, furthermore, the effect of buildings and other urban structures on

pollutant diffusion and accumulation patterns (Vardoulakis et al, 2003). This study

makes use of a pollution dispersion model and involves its application and

validation within a street canyon in a built-up area of Leeds.

1.2 Objectives and Scope of the Study

The central objective of this dissertation is to study the dispersion of air pollution

within an urban street canyon and, specifically, along a section of Otley Road in the

area of Headingley, in northern Leeds (Fig. 1.1).

The two main aims are to monitor and to model the air quality in this urban

environment and, thus, evaluate the performance of a modern dispersion model. The

first objective is to collect measurements of air pollutant concentrations from motor

vehicles in the examined road as well as the associated traffic and meteorological

data. Subsequently, the goal is to model the dispersion of the most important

pollutants in the street canyon using the street canyon module of an advanced air

quality model.

Finally, through the comparison of the monitored and the predicted pollution levels,

the applied model can be validated and the presence of street canyon effect on the

pollutants’ dispersion can also be demonstrated.

The scope of this project is intended to include monitored data collection and

dispersion modelling techniques in an urban street canyon in order to examine and

interrelate a wide range of physical and environmental variables.


- 3 -

Figure 1.1 The location of the examined street canyon in Headingley, Leeds.

Source: Google Maps (2006)

Due to the lack of time and resources, the monitoring campaign will be limited to

concentrations of Carbon Monoxide (CO) across the selected urban area. The

installed equipment by the Institute for Transport Studies (ITS), University of Leeds,

can measure CO concentrations for adequate period of time along the examined

street canyon and at a background location. Meteorological and other useful

background pollution data have been provided by the Leeds City Council. The

required traffic data have been collected from relevant ITS projects and from manual

classified vehicle counts in the area.

Regarding the modelling procedure, it is possible to study a wider range of

pollutants and scenarios in the area. Hence, different meteorological and topographic

Street Canyon


- 4 -

characteristics can be analysed. For instance, various wind speeds and directions or

receptor locations can be investigated and the effects on pollutants dispersion could

thus be deduced. In this study a recently updated air quality model is used for the

prediction of pollution levels within the examined urban area. The Windows based

Operational Street Pollution Model (WinOSPM, Version 5.0.64) developed by the

National Environmental Research Institute of Denmark (NERI, 2006) is believed to be

a well accepted and widely applied model that is capable of assessing air quality in

confined urban environments.

1.3 Structure of the Dissertation

The structure of this project is outlined below:

The second chapter contains the literature review on traffic pollution and, in

particular, a brief description of the main air pollutants and the legislative

framework for air quality standards in the United Kingdom (UK) and European

Union (EU). The emission and dispersion of air pollution are also examined giving

emphasis to the behaviour of the gases in constrained urban areas and street canyons.

Moreover, the process of monitoring traffic pollution is reviewed along with the UK

National Air Quality Strategy. The chapter closes with a description of the main

types of air quality models and, in particular, the OSPM dispersion model, which has

been used in this study of pollution dispersion in urban street canyons.

The methodology of the study is developed in Chapter 3. The first part includes the

experimental method of the monitoring of the pollutants examined here. The selected

site and the required equipment are described as well as the process of data

collection, i.e. the meteorological data, the manual and automatic traffic counts and

the in situ and background measurements of the pollutant concentrations. In the

second section, the setting and the application of the air quality model is presented

and the input and output data are also analysed.


- 5 -

Chapter 4 includes the results of the monitoring and modelling procedures. The

observed data from the monitoring campaign are first analysed. The predictions of

the AQ model are then compared against the measured concentrations in order to

evaluate the performance of OSPM in the current situation.

Finally, the fifth chapter summarises the main conclusions of the study regarding the

performance and the validation of the applied dispersion model. Recommendations

for future work and applications are also mentioned at the end.

- 6 -

Chapter 2

Traffic Pollution

Although good quality of life and robust level of economy are strongly correlated with

an efficient and flexible transport system, current situation shows that the radical

increase in road transport causes an important and growing threat to natural

environment and human health. The transport sector is the fastest growing consumer of

energy and producer of greenhouse gases in Europe, jeopardising thus the EU reaching

the 2010 emissions reduction target set under the Kyoto protocol. Vehicle technology

and fuel improvements have resulted in noticeable decreases of emissions of particular

pollutants. Nevertheless, air quality in most European cities is still poor and, hence,

transport policies ought to aim at the control of traffic growth and the promotion of

sustainable transport modes (EEA, 2006a).

The risk of human exposure to high pollutant concentrations is significantly increased in

densely populated urban locales. Air quality limit values are set in order to care for

public health, but they are repeatedly exceeded, especially in busy roads and other

urban hotspots. The pollutants emission and dispersion modelling is a powerful tool for

determining which pollutants’ reductions are needed in certain areas, so that they

remain below the predefined emission ceilings. The analysis of air quality scenario

projections at street level and the impacts of particular policies and measures are

possible with the use of street scale models. Thorough local traffic data, air pollution

measurements, meteorological and topographical data of the problematic sites are

I.A. Karousos Modelling of Air Quality in Street Canyons 2. Traffic Pollution

- 7 -

normally required as inputs to these models in order to obtain accurate air quality

prognoses at street level (EEA, 2006b).

This chapter begins with the main components of air pollution and the legislative

framework of air quality control. Subsequently, the mechanisms of pollutants’ emission

and dispersion are reviewed, laying emphasis on urban areas and street canyons, as

well as the principles of relevant monitoring and modelling techniques.

2.1 Main types of air pollutants

The most important air pollutants, which are related to motorised traffic and mostly

concern local authorities nowadays, are carbon monoxide, nitrogen oxides,

hydrocarbons and particulate matters. The UK Department for Environment, Food and

Rural Affairs (DEFRA, 2006) has listed the main air pollutants, their sources and impacts

on human health as well as some characteristic time-series data and trends of emissions,

as summarised below. These substances are mostly produced by fossil fuel burning and

in Table 2.1, they have been categorised in terms of contribution to greenhouse

phenomenon and acidification, toxicity and involvement in Local Air Quality strategies.

Carbon monoxide (CO) is a toxic gas produced by combustion processes and by

the oxidation of hydrocarbons and other organic compounds. It reduces the

capacity of the blood to transport oxygen and deliver it to the tissues. Figure 2.1

illustrates the trends in CO emissions by emission source from 1970 to 2004. The

introduction of catalytic converters on petrol vehicles and, to some extent, the

increasing numbers of diesel cars are the main reasons for the dramatic reduction

of total emissions after 1990. The diagram in Figure 2.2 shows that the prevailing

vehicle type has logically been private car and consists, thus, the predominant

source of traffic-related CO pollution.


- 8 -

Table 2.1 Air pollutant types (including greenhouse gases).

Pollutant Greenhouse gas

Acid gas

Ozone precursor1

Toxic pollutant2

Local Air Quality

Carbon dioxide CO2 x Methane CH4 x x Nitrous oxide N2O x Hydrofluorocarbons HFC x x Perfluorocarbons PFC x x Sulphur hexafluoride SF6 x Nitrogen oxides NOx (NO2 + NO) indirect x x x

Sulphur dioxide SO2 indirect x x Particulates PM10 x Black smoke BS x Carbon monoxide CO x x Ozone O3 x Non-methane volatile organic compounds NMVOC indirect x x

Benzene x 1,3 butadiene x Ammonia NH3 x Hydrogen chloride HCl x Hydrogen fluoride HF x Arsenic As x Cadmium Cd x Chromium Cr x Copper Cu x Mercury Hg x Nickel Ni x Lead Pb x x Selenium Se x Vanadium V x Zinc Zn x Persistent organic pollutants POPs x 1 Ozone is produced by photochemical reactions involving VOCs & NOx in the lower atmosphere. 2 Includes heavy metals & POPs. Source: Department for Environment, Food and Rural Affairs (DEFRA, 2006)


- 9 -

0

2

4

6

8

10

12

14

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

CO

(106 to

nnes

)

Other

Residential

Road transport

Figure 2.1 Estimated CO emissions by source. UK, 1970-2004.

(adapted from DEFRA, 2006)

0

2000

4000

6000

8000

10000

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

CO

(103 to

nnes

)

Passenger cars LGVs HGVs Buses Mopeds & Motorcycles

Figure 2.2 Estimated CO emissions by vehicle type. UK, 1970-2004. (adapted from DEFRA, 2006)


- 10 -

Nitrogen oxides (NOx) are acid gases and ozone precursors contributing largely to local

air pollution. The emissions can have negative effect on human health and vegetation.

Nitrogen dioxide (NO2) is believed to affect airways and lung function both acutely and

chronically, particularly in people with asthma. Figure 2.3 shows estimated emissions of

NOx by source category for the period 1970-2004. Total emissions reached their

maximum in late 1980s, but then considerably declined between 1990 and 2004, mainly

due to the use of three way catalysts for petrol cars and also cleaner operation of major

combustion plants. The shift from some non-catalyst petrol to diesel cars has also played

a smaller part.

0

500

1000

1500

2000

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

NO

x (1

03 tonn

es)

Passenger cars LGVs HGVs Buses Mopeds & Motorcycles*

* Approximately 1,000 tonnes per year

Figure 2.3 Estimated NOx emissions by vehicle type. UK, 1970-2004. (adapted from DEFRA, 2006)

Ground level ozone (O3) is a natural atmospheric component, but its

concentration can rise, when reactions between NOx, oxygen and Volatile


- 11 -

Organic Compounds (VOCs) occur in the presence of sunlight. The formation of

ozone as a secondary pollutant is widely affected by the weather and episodes of

high levels usually occur during summer due to the prolonged sunshine, high

temperatures (>20°C) and calm winds. Ground level O3 can persist for several

days and can be transferred long distances. Ozone concentrations are generally

higher at rural sites than in urban areas, because nitric oxide (NO), which is quite

common in cities, can react with and deplete O3 to from nitrogen dioxide (NO2).

Possible effects on human health include eyes and nose irritation and even

damage of the airway lining in case of exposure to high levels.

Airborne Particulate Matters (PM) is a very diverse category of air pollutants,

including microscopic compounds with various physical and chemical properties

and of many forms. Apart from motor traffic, the particles can be formed from

power generation, construction work, gas-to-particle conversion and other

photochemical reactions. Road transport in the UK now contributes 23% of all

PM10 emissions with the main sources being exhaust of diesel vehicles,

automobile tyre and brake wear, and resuspension of road dust and soil particles.

Figure 2.4 shows trends in PM10 emissions by vehicle type for the last decades.

The drastic reduction since 1990 is generally due to modern low-emission engines

and particulate traps installed in heavy duty vehicles. The airborne particles size

plays important role in their behaviour and the potential hazards against human

health. Fine particles, 10μm in diameter or smaller, can penetrate deep into the

lungs and their accumulation might cause premature deceases among people

with pre-existing lung and heart disease. At present, ultrafine (<0.1 μm) and

nanoparticles (<0.05 μm) constitute a growing field of research, as they might

consist the main fraction of modern engine emissions and their environmental

and health impacts may be stronger than that of fine particles (Le Bihan et al.,

2004).


- 12 -

0

20

40

60

80

100

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

PM10

(103 to

nnes

)

Passenger cars LGVs HGVs Buses Mopeds & motorcycles Vehicle tyre & brake wear

Figure 2.4 Estimated PM10 Emissions by vehicle type. UK, 1970-2004. (adapted from DEFRA, 2006)

Volatile Organic Compounds (VOCs) are ozone precursors and include a wide

variety of chemical compounds, such as hydrocarbons (alkanes, alkenes,

aromatics), oxygenates (alcohols, aldehydes, ketones, ethers) and halogen

containing species. The notable decrease of road transport VOC emissions since

1990 is mainly due to the introduction of catalytic converters for petrol cars and,

to some degree, to replacement of non-catalyst petrol cars with diesel vehicles.

Except for fossil fuels burning, a significant proportion of VOCs come from

sources, such as motor fuel and solvent evaporation, refining of petrol and other

industrial processes, and natural sources. The most important environmental

impact of non-methane VOCs relates to their contribution to ground level O3

formation, but they may also cause various health effects. Most VOCs are non-

toxic, but others like benzene and 1,3-butadiene are carcinogenic and have

damaging effects on human health.


- 13 -

Sulphur dioxide (SO2) is an acid gas and is principally produced by fossil-

fuelled power stations, but also from fuel use in manufacturing industries and

construction. Total SO2 emissions in the UK fell dramatically between 1990 and

2004. Emissions from power stations declined by 82%, largely as a result of a

reduction in coal use and introduction of outlet gas desulphurisation plants.

Emissions from motorised traffic have decreased by 85% since 1998 due to

reduction in the sulphur content of fuel. Sulphur dioxide can have negative

effects on human health and particularly the lining of the nose, throat and

airways of the lung, and among people with asthma and chronic lung disease.

Among the main air pollutants belong also: (i) Ammonia, which causes nitrogen

enrichment and potentially acidification, and may also contribute to the creation

of particulate matters through atmospheric chemistry. (ii) Hydrogen chloride

and hydrogen fluoride, which are both acid gases. (iii) Lead and other heavy

metals, which can cause, among a range of health problems, deterioration of the

immune, the metabolic and the nervous system, and many are known or

suspected carcinogenic substances. (iv) Persistent organic pollutants, which

include: polycyclic aromatic hydrocarbons, polychlorinated biphenyls, dioxins

and furans; trace quantities can be found in all areas of the environment and they

have varying levels of toxicity and an accumulative effect in humans and

vegetation.

2.2 National Air Quality Strategy

This section incorporates the general legislative framework regarding vehicle emissions

and the acceptable limit values for pollution levels. As illustrated above, emissions and

concentrations of most airborne pollutants have declined extensively as a consequence

of the implementation of national and local measures in UK.


- 14 -

According to the 1995 Environment Act, the UK Government and the devolved

administrations for Scotland and Wales are responsible for policy and legislative issues

related to the environment and air quality. The transboundary nature of many air

pollutants, however, encouraged the formation of a UK-wide air quality strategy with

common aims for all parts. The 2000 National Air Quality Strategy (NAQS) for England,

Scotland, Wales and Northern Ireland was produced on that basis and included

standards and objectives for improving ambient air quality. The Act also initiated the

concept of local air quality management (LAQM), where local authorities are required

regularly to review and evaluate the current and future pollution levels in their areas

against the prescribed targets in the Strategy (DEFRA, 2003).

The National Air Quality Strategy contains the governmental policies for reducing air

pollution and introduces standards for certain air pollutants, which are set to be

achieved by certain target dates. These pollutants were revised in the NAQS Addendum

(published in 2003) and now comprise of: sulphur dioxide (SO2), fine particles (PM10),

nitrogen dioxide (NO2), carbon monoxide (CO), lead (Pb), benzene, 1,3-butadiene,

ground level ozone (O3) and polycyclic aromatic hydrocarbons (PAHs). The updated

objectives are summarised in Table 2.2 (DEFRA, 2006).

The main goals of this initiative as outlined in the updated NAQS Addendum (DEFRA,

2003) are to:

map out future air quality policy in the UK in the medium term and to the

greatest possible extent;

provide realistic protection to human health by setting health-based limits for air

pollutants;

set objectives to protect the natural environment, i.e. vegetation and ecosystems;

describe present and future levels of air pollution; and

provide a framework to help recognise what can be done at local, national and

international level to improve air quality.


- 15 -

Table 2.2 New air quality objectives included in the NAQS for protecting human health.

Pollutant Objective Concentration(1)

measured as: To be

achieved by 16.25μg/m3 (5ppb)

in England & Wales running annual mean 31/12/2003

5μg/m3 (1.54 ppb) in England & Wales annual average 31/12/2010 Benzene

3.25μg/m3 (1 ppb) in Scotland & N. Ireland running annual mean 31/12/2010

1,3-Butadiene 2.25μg/m3 (1ppb) running annual mean 31/12/2003 10 mg/m3 (8.6 ppm)

in England, Wales & N. Ireland (2) maximum daily

running 8-hour mean 31/12/2003 Carbon monoxide

10 mg/m3 (8.6 ppm) in Scotland running 8-hour mean 31/12/2003

0.5μg/m3 annual mean 31/12/2004 Lead

0.25μg/m3 annual mean 31/12/2008 200μg/m3 (105ppb)

not to be exceeded more than 18 times a year (3)

1-hour mean 31/12/2005 Nitrogen dioxide

40μg/m3 (21ppb) annual mean 31/12/2005

Ozone 100μg/m3 (50ppb)

not to be exceeded more than 10 times a year (3)

maximum daily running 8-hour mean 31/12/2005

Polycyclic aromatic hydrocarbons 0.25ng/m3 B[a]P (3),(5) annual average 31/12/2010

266μg/m3 (100ppb) not to be exceeded more than 35

times a year 15-minutes mean 31/12/2005


times a year 1-hour mean 31/12/2004 Sulphur dioxide


times a year 24-hour mean 31/12/2004

50μg/m3 not to be exceeded more than 7 times a year in UK (apart

from London) & 10 times a year in Greater London (3),(4)

24-hour mean 31/12/2010

20μg/m3 in England (apart from London), Wales & N. Ireland (3),(4) annual mean 31/12/2010

18μg/m3 in Scotland (3),(4) annual mean 31/12/2010

Particles

23μg/m3 in Greater London (3),(4),(6) annual mean 31/12/2010 (1) Conversions of ppb and ppm to µg/m3 and mg/m3 at

20oC and 1013mb (2) During 2002 the original strategy objective for CO was

replaced by a more stringent one (3) Objectives are provisional (4) Measurements are in gravimetric units

(5) N. Ireland adopted the same PAHs objective as England, Wales and Scotland in April 2004

(6) It is proposed that London should work towards a 20µg/m3 annual mean target after 2010, with the aim of achieving it by 2015, where cost effective and proportionate local action can be identified.

Source: DEFRA, 2003 & 2006


- 16 -

2.3 Vehicle emissions

Vehicle emissions are mainly originated from the fuel burning engine operation, which

transmits to the tail pipe a variety of elements and chemical compounds. The primary

ones are: carbon dioxide, a mixture of NOx (of which more than 90% is in the form of

NO and the rest is NO2), carbon monoxide from incomplete burning, as well as multiple

unburned and chemically transformed hydrocarbons, such as benzene and 1,3-

Butadiene, methane (CH4) and other alkanes, and other complex polymeric aromatics.

Additional products are particulate matters of different size and composition, mainly of

condensed carbon material, emitted by diesel and by poorly maintained petrol vehicles.

Road traffic emissions also include other processes, such as direct evaporation of volatile

fuel components and leaks from fuel tank, as well as dust particles from tyre and brake

wear and from re-suspended material of road surface erosion. (Vardoulakis et al., 2003;

Ropkins, 2006).

Such traffic-related emissions were considerably high in the last decades due to the vast

growth of road transport. Motorised traffic has, thus, been converted into the primary

source of air pollution, when studying congested urban networks or industrialised

regions. On the other hand, at the same time the air pollution emitted from

manufacturing and domestic sources has generally declined to a great degree due to the

various governmental Acts and the stricter regulations introduced in the majority of the

developed countries (Sharma and Khare, 2001).

As shown earlier though, emission levels of main pollutants have been reduced

significantly over the recent years mostly as a result of the stricter vehicle emission and

fuel quality EURO standards and in spite of the further growth of motor traffic (15%

since 1990). The graph in Figure 2.5 is illustrative of the current atmospheric emission

levels in the UK and of the mitigated share of traffic induced emissions. Technological

innovations can contribute to further emissions reduction over the long-term, but the


- 17 -

large proportion of older vehicles that remain on the road networks and the increasing

trends of traffic volumes are likely to have an adverse effect on such improvement

(DEFRA, 2006).

0

500

1000

1500

2000

2500

3000

3500

CO NOx NMVOC* PM10

103 to

nnes

Total

Road transport

*Non-methane VOCs incl. benzene and 1,3-butadiene.

Figure 2.5 Total and traffic-related emissions of key air pollutants in UK for 2004.

(Source: NETCEN & Office for National Statistics; ONS, 2006)

The basic principle in the emissions calculation method is that the core vehicle emissions

comprise of the various exhaust gases and the VOCs produced by fuel evaporation. An

additional fundamental assumption is that engine emissions are higher when an engine

is started, because it runs below its normal operating temperature and the fuel is used

inefficiently. The total emissions are, thus, the sum of the hot engine and cold start

emissions plus the possible evaporative losses (EC-MEET, 1999):

ETOTAL = EHOT + ECOLD + EEVAPORATIVE

Each of these components is, however, affected by several parameters and modification

factors:

The vehicle type, model and speed, the engine size and pollution control, the age,

mileage and maintenance level of the vehicles are important factors specifying

the quantity and quality of the emissions. The passenger cars versus heavy duty


- 18 -

vehicles or the use of three-way catalytic converter in the exhaust system are

relevant classification criteria.

The type and grade of fuel used, e.g. petrol, diesel, LPG or biodiesel, and

standard versus high octane fuel, also play important role in engine emissions.

The driving activity includes: urban, rural or motorway trips (i.e. average speed,

congestion, stops/starts, trip length); driver’s behaviour (e.g. normal, passive or

aggressive driving); high or low engine load demand (driving uphill or downhill,

vehicle load or use of in-car air conditioning system); and ‘cold start’ versus

‘warm/hot’ vehicle operation. Each of these variables can differentiate the

amount of emissions from the same vehicle.

External factors, such as road condition (quality of road surface, road gradient

etc.) and ambient air temperature, pressure, humidity and other environmental

parameters also determine the exhaust emissions (Ropkins, 2006; Namdeo, 1995).

The calculation of vehicle emissions is clearly a complex and multipart procedure. In

general, exhaust gas emissions are the most easy to monitor and, thus, to accurately

model. On the other hand, pollutants emitted from other points of the vehicle (e.g.

vaporisation of fuel spirit during refuelling, diurnal breathing, hot soak and running

losses) or from the passage of vehicles (brake/tyre/road-surface wear, re-suspended

road dust etc.), cannot be easily quantified and they are, hence, poorly represented in

emissions modelling.

Furthermore, the above mentioned parameters that can be quantified or accurately

estimated, are only a small proportion of all the possible combinations of factors that

influence vehicle emissions. Most of the emission models use different emission factors1

for only certain conditions and apply appropriate quantifiable correction factors to the

basic calculations (Ropkins, 2006).

1 Emission factor is the amount of emission produced in a given period under certain conditions divided by that time period or the by distance travelled/fuel consumed during that period (Ropkins, 2006).


- 19 -

A major category of emission models contains those that utilise average speed-emission

relationships to estimate emissions during a trip. The European COPERT III (COmputer

Programme to calculate Emissions from Road Transport III) is a widely used model,

which calculates total emissions by combining driving activity data for each vehicle

class with emission factors suitable for each driving situation, climatic conditions etc. An

outline of the parameters required and the intermediate computed values is provided in

the flow chart of Figure 2.6. The varying driving conditions play an important role, as

they influence engine operation conditions and, thus, cause divergent emission profiles.

Hence, the erratic driving and emission performance in each trip is further represented

by distinguishing urban, rural and motorway driving situations. Each category has

different average speeds and, thus, differing activity data and emission factors. The cold

start emissions are calculated additionally to the ‘hot emissions’, which occur after the

engine and the catalyst have warmed-up. The impact of cold start over-emissions is

mainly attributed to urban driving, because the large majority of trips are assumed to

start within urban areas contributing, thus, to higher pollution levels there

(Ntziachristos and Samaras, 2000). When calculating emissions from urban driving, it is

also important to consider the effect of driver behaviour, which can be quantified by

dividing the modelled journey according to driving mode, i.e. idling, accelerating,

constant-speed cruising and decelerating. The associated emission factors for each

operating mode are different. This modal method can show the effect of traffic

conditions, particularly at congested intersections, as well as the impacts of traffic

management schemes (e.g. traffic signal coordination).

Instantaneous emission models are another key tool used to estimate vehicle emissions.

Measuring campaigns of instantaneous emission data are quite rare and elaborate, so by

use of these models it is possible to predict second-by-second exhaust emissions for a

variety of vehicle types, engine technologies and emission control devices. Such models

can accurately estimate emissions during the acceleration mode, which is the period

with the highest emission rates for modern engine vehicles. However, they require a


- 20 -

large amount of input data and the

appropriate vehicle and engine

operation parameters must be

specified or calculated in advance

(Tate, 2006).

Figure 2.6 Flow chart of the application of

COPERT-III baseline methodology. (Source: Ntziachristos and Samaras, 2000)

2.4 Dispersion of air pollution in urban areas

Having estimated the vehicle emissions, the next step is to examine the pollutants

dispersion in order to calculate their concentrations across the examined area. Air

pollution problems have typically occurred at local scale, i.e. in the surroundings of

sporadic point or area sources. More recently, environmental policy has had to address

also global scale problems, such as the greenhouse effect, climate change, ozone

depletion etc. Table 2.3 contains some key policy issues related to the atmospheric

environment including, among others, acidification, photo-oxidant formation, air toxics

and the modern problem of urban air pollution.


- 21 -

Table 2.3 Environmental policy issues and analogous scales of dispersion phenomena.

Scale of dispersion phenomenon Policy issue

Global Regional-to-continental

Local-to-regional Local

Climate change X

Ozone depletion X X

Tropospheric ozone X

Tropospheric change X

Acidification X

Nutrification X

Summer smog X X

Winter smog X X

Air toxics X X X

Urban air quality X

Industrial pollutants X X

Nuclear emergencies X X X

Chemical emergencies X X X

Source: Moussiopoulos et al., 1999

The dispersion of air pollution largely depends on processes in the atmosphere which

are differentiated with regard to their spatial and temporal scale. These scales extend

from macroscale (typical lengths >1,000km), including global and most of regional-to-

continental phenomena, where the air flow is largely related to synoptic phenomena

(e.g. the geographical distribution of pressure systems) to microscale (typical lengths

<1km), where atmospheric flow is very complex and depends strongly on the detailed

surface features. Local scale dispersion incidents belong to microscale and can be well

described in appropriate simulation modelling tools (Moussiopoulos et al., 1999).

In such models the factors that typically have to be considered are: (i) the shape and size

of buildings and other obstacles, (ii) their position with regard to the wind direction, (iii)

the variability in wind speeds and directions, (iv) the amount of mechanical turbulence


- 22 -

mainly dependent on the vehicle

induced mixing and the

formation of air vortices because

of the surface roughness (see Fig.

2.7), (iv) the thermal turbulence

effect, which is determined by the

temperature structure of the

lower atmosphere and relates to

the vertical convection of heat.

Street Canyons

An interesting microscale dispersion phenomenon occurs at urban street canyons, which

are ideally narrow streets between buildings arrayed continuously along both sides.

This distinctive configuration impedes good ventilation when wind is perpendicular to

the street axis and, thus, results in poorer dispersion and often higher pollutant

concentrations than current air quality standards (Nicholson, 1975).

The cross-section, illustrated in Figure 2.8, is of a typical urban street canyon flanked by

buildings on both sides. In reality, a wide range of geometries can be found with

varying structure heights and shapes along the street, with gaps between the buildings

or even with buildings only on one side. In this case, the effect of a regular, symmetric

street canyon is showed, with an aspect ratio approximately equal to one1 and buildings

of similar height on either side. When the above roof-level wind direction is

perpendicular to the street, the wind flow skims over the canyon producing a primary

air vortex between the buildings, which is the main dispersion mechanism for the air

pollutants. The diffusion is generally affected by the rate of the fresh air exchange

1 i.e. Average Buildings Height/Average Street Width ≈ 1

Figure 2.7 Perpendicular flow systems in urban areas for different road widths.

(adapted from Vardoulakis et al., 2003)


- 23 -

between the street and its outlets, i.e. the above roof level and the connecting roads. The

recirculating wind flow cannot remove effectively the street-level pollutants and, in

addition, it contributes to the formation of substantial pollution gradient across the

street canyon, as also verified by field measurements. The leeward side of the canyon

(see Fig. 2.8) shows overall higher concentrations than the windward due to the

transferred pollutants by the main wind vortex. Secondary weaker vortices can be

created in the bottom side corners of the canyon or in small cavities causing, thus,

localised pollution hotspots. The strength of the canyon vortices largely depends on the

speed of the roof-level wind, but it is also affected by the vehicle induced mechanical

and the thermal turbulence or the reflection off the various obstacles within the street

(e.g. trees, kiosks, balconies etc.).

More complex wind flows occur near the ends of the canyon at intersections with other

streets. There, the low-pressure corners and the wind circulation create horizontal air

vortices which bring fresh air into the canyon. This ventilation mechanism, however,

becomes weaker as the street length increases. Similarly complex flow channelling

effects occur near gaps

between the canyon buildings

and also within asymmetric

canyons, where the down-

wind and up-wind buildings

have different heights

(Vardoulakis et al., 2003).

Figure 2.8 Vertical cross-section of a typical symmetric urban street canyon. The recirculating wind flow is shown in the case of perpendicular roof level wind.

(Source: NERI, 2006)


- 24 -

2.5 Monitoring traffic pollution

The objectives of air quality strategy are based on constantly updated air pollution

measurements from the national air pollution monitoring network and the most up-to-

date modelling tools of air quality level predictions. More than 1600 national AQ

monitoring sites are located around the UK. They are divided into automatic and non-

automatic network monitors and collect several types of data depending on the local

requirements. Most recent developments include the dissemination to the public of

hourly collected data from approximately 120 automatic monitoring points, the

operation of sites, which are capable of collecting detailed information on ultra-fine

particles, as well as an increased number of monitoring sites of PAHs (DEFRA, 2003).

Nevertheless, the amount of permanent air quality monitoring stations in a city is

practically limited because of the initial and operational costs, the size and shape of the

equipment, the power supply possibility etc. It is, hence, important to ensure the

optimal utilisation of the available monitoring equipment and of the obtained data,

which can be supplemented via alternative measurements and modelling techniques in

order to fully assess air quality and population exposure in dense built-up areas.

According to the EU legislation on air quality, monitoring stations should be sited where

the highest concentrations and human exposure risks occur. Confined urban

environments should be avoided, so that the receptor measurements can represent the

air pollution levels in a surrounding area of at least 200m2. Other guidance requires that

the height of the monitoring point is between 1.5m (i.e. human breathing zone) and 4m,

no less than 25m away from major junctions and than 4m from the middle of the nearest

traffic lane. For NO2 and CO, the sampling inlet should be less than 5m from the

kerbside, and for PM10 and benzene it should be placed near the building façade (but no

less than 0.5m from the nearest wall) (Vardoulakis et al., 2005).


- 25 -

Several techniques have been developed for monitoring air pollutants (e.g. continuous

monitoring using standard gas analysers or diffusive and pumped sampling using tubes

filled with an appropriate adsorbent, etc.) and particles (e.g. filtration and impaction),

each of them with advantages and disadvantages that make it proper or not for a

specific use. The response time (i.e. the time interval in which the sample is taken), is a

key criterion for the suitability of the selected method. Standard gas analysers are

sensitive and fast enough for real time measurements of CO, NOx and O3

concentrations. Appropriate averaged results can then be produced over short time

periods so that they are comparable with the regulatory standards. On the other hand,

diffusive samplers have a fairly long response time (e.g. one/two days to four weeks),

which makes them preferable when sampling substances with cumulative effects on

human health (e.g. benzene) or for spatial variation measurements, air quality mapping

and personal exposure studies, since they are portable devices (Vardoulakis et al., 2003).

Having compared monitoring data from two streets in Copenhagen, Berkowicz et al.

(1996) demonstrated that roadside measurements are highly site dependent, even within

the same street. Other studies have also shown the dependence on local wind flows and

the interaction with street and buildings geometry. So, it is very important to focus on

the ability of sampling points to characterise traffic pollution in complex urban areas

and also to be combined with modelling tools. The ideal urban air quality management

would avoid costly monitoring campaigns as well as excessive dependence on models,

by maximising the representativeness of AQ monitoring stations and following the

existing location criteria for roadside receptors (Vardoulakis et al., 2005).

Vardoulakis et al. (2003) suggest two main categories of permanent AQ stations in a city:

(a) the roadside and (b) the urban background stations. Roadside stations are usually

located on the pavement of busy streets, avenues or junctions (as described earlier),

whereas background stations are placed in parks or other urban locations away from

road traffic.


- 26 -

2.6 Modelling traffic pollution

Several modelling studies, related to the effect of buildings and other urban structures

on pollutant concentrations and dissipation patterns, were driven by the increasing need

for research on the impacts of air pollution in urban environments. Dispersion models

are now extensively used for assessing roadside air quality by predicting current and

future pollution levels, as well as their temporal and spatial variations. When used in a

knowledgeable way, they are able to provide most detailed information about the

physical and chemical processes that govern the diffusion and transformation of

atmospheric substances (Vardoulakis et al., 2003).

2.6.1 Types of air quality models

Models describing the dispersion of air pollution in the atmosphere can be

distinguished in many ways, such as: their spatial (global, regional-to-continental, local-

to-regional or local) and temporal scale (episodic or statistical, long-term models), the

treatment of the transport equations (Eulerian, Lagrangian models etc.) and of the

different processes (chemistry, wet and dry deposition). Table 2.4 summarises the main

existing model types and their brief descriptions.

Since late 1950’s atmospheric dispersion models based on Gaussian distribution and

Pasquill-Gifford classes have been used for legislative reasons in Europe. During the last

two decades dispersion models have been developed based on boundary layer

parameterisation occurred due to increasing understanding of both the structure of the

boundary layer and the dispersion science. The meteorological input data to these

models are produced via new methods where the vertical profiles of speed, temperature

and turbulence are dependent on the height of the boundary layer and a Monin-

Obukhov length scale determined by the temperature, the friction velocity and the heat


- 27 -

flux. More reliable and accurate modelling results are required due to the expanding

utilisation of practical operational models for regulatory and planning purposes. The air

quality guidelines are becoming increasingly stricter and detailed in many countries and

evaluated models are required to meet the needs of modern air quality management,

especially in urban areas (Moussiopoulos et al., 1996).

Table 2.4 Main types of existing air quality models.

Model type Description

Plume-rise models

In most cases, pollutants injected into ambient air possess a higher temperature than the surrounding air. Most industrial pollutants, moreover, are emitted from stacks or chimneys and possessing, thus, an initial vertical momentum. Both factors (thermal buoyancy and vertical momentum) contribute to increasing the average height of the plume above that of the smokestack. Plume-rise models calculate the vertical displacement and general behaviour of the plume in this initial dispersion phase. Both semi-empirical and advanced plume-rise formulations are available.

Gaussian models

The Gaussian plume model is the most common air pollution model. It is based on the assumption that the plume concentration, at each downwind distance, has independent Gaussian distributions both in the horizontal and vertical direction. Almost all the models recommended by the U.S. Environmental Protection Agency are Gaussian. Gaussian models have been modified to incorporate special dispersion cases. A simplified version of Gaussian model, the Gaussian climatological model, can be used to calculate long-term averages (e.g. annual values).

Semi-empirical models

This category consists of several types of models which were developed mainly for practical applications. In spite of considerable conceptual differences within the category, all these models are characterised by drastic simplifications and a high degree of empirical parameterisations. Among the members of this model category are box models and various kinds of parametric models.

Eulerian models

The transport of inert air pollutants may be conveniently simulated by the aid of models which solve numerically the atmospheric diffusion equation, i.e. the equation for conservation of mass of the pollutant (Eulerian approach). Such models are usually embedded in prognostic meteorological models. Advanced Eulerian models include refined sub-models for the description of turbulence (e.g. second-order closure models and large-eddy simulation models).

Lagrangian models

As an alternative to Eulerian models, the Lagrangian approach consists in describing fluid elements that follow the instantaneous flow. They include all models in which plumes are broken up into elements such as segments, puffs, or particles. Lagrangian models use a certain number of


- 28 -

fictitious particles to simulate the dynamics of a selected physical parameter. Particle motion can be produced by both deterministic velocities and semi-random pseudo-velocities generated using Monte Carlo techniques. Hence, transport caused by both the average wind and the turbulent terms due to wind fluctuations is taken into account.

Chemical modules

Several air pollution models include modules for the calculation of chemical transformation. The complexity of these modules ranges from those including a simple, first-order reaction (e.g. transformation of sulphur dioxide into sulphates) to those describing complex photochemical reactions. Several reaction schemes have been proposed for simulating the dynamics of interacting chemical species. These schemes have been implemented into both Lagrangian and Eulerian photochemical models. In Eulerian photochemical models, a three-dimensional grid is superimposed to cover the entire computational domain, and all chemical reactions are simulated in each cell at each time step. In the Lagrangian photochemical models a single cell (or a column of cells or a wall of cells) is advected according to the main wind in a way that allows the injection of the emission encountered along the cell trajectory.

Receptor models

In contrast to dispersion models (which compute the contribution of a source to a receptor in effect as the product of the emission rate multiplied by a dispersion coefficient), receptor models start with observed concentrations at a receptor and seek to apportion the observed concentrations at a sampling point among several source types. This is done based on the known chemical composition of source and receptor materials. Receptor models are based on mass-balance equations and are intrinsically statistical in the sense that they do not include a deterministic relationship between emissions and concentrations. However, mixed dispersion-receptor modelling methodologies have been developed and are very promising.

Stochastic models

Stochastic models are based on statistical or semi-empirical techniques to analyse trends, periodicities, and interrelationships of air quality and atmospheric measurements and to forecast the evolution of pollution episodes. Several techniques are used to achieve this goal, e.g. frequency distribution analysis, time-series analysis, Box-Jenkins and other models, spectral analysis, etc. Stochastic models are intrinsically limited because they do not establish cause-effect relationships. However, statistical models are very useful in situations such as real-time short-term forecasting, where the information available from measured trends in concentration is generally more relevant (for immediate forecasting purposes) than that obtained from deterministic analyses.

Source: Zanetti, 1993

Regarding the requirements for input data into the modelling tools, there are generally

two main categories of information required (Moussiopoulos et al., 1996):


- 29 -

Emissions. For the traffic models, typical descriptive information for the road

network work are required as input data. This could be number of cars per day,

number of lanes, average driving speed, road gradient and explanation of the

surroundings, i.e. open terrain, scattered buildings or street canyon.

Meteorology. The screening type models use default meteorology data as input,

describing a critical meteorological situation. Other models use a set of

meteorological cases as input, described in wind and stability classes. The new

types of models make use of pre-processed meteorological data based on

similarity theory for the atmospheric surface layer. The measurements that are

required include variables, such as wind and temperature profiles, cloud cover or

solar radiation, surface roughness etc.

The Operational Street Pollution Model

The Operational Street Pollution Model (OSPM) was developed for supplementation to

standard monitoring activities and for assessment of abatement strategies at the

National Environmental Research Institute (NERI) of Denmark. The model contains a

simplified description of flow and dispersion conditions in urban roads. Concentrations

of exhaust emissions are calculated using a combination of a plume model for the direct

contribution and a box model for the recirculating pollution part in the street. Despite

the simplified parameterisation, OSPM is capable of a proper simulation of the

dependence of air pollution levels on meteorological conditions, such as wind speed

and wind direction. A recent improvement is the modelling of turbulence in the street

by taking into consideration the effect of atmospheric turbulence due to wind velocity,

but also due to vehicle induced mixing, which dominates for low and calm wind cases.

The model includes also a chemical sub-model which is used to calculate the

transformation of NO to NO2. Using actual meteorological data and estimations of

emissions, as well as a priori assumptions regarding flow and dispersion characteristics,

the model provides hourly values of concentrations at predefined receptor locations in

the examined street (Gokhale et al., 2005).

Chapter 3

Study Methodology

This air quality study is basically divided into two sections, the monitoring and the

modelling part. This chapter contains both the experimental and the computational

methodology used to investigate the street canyon effect and evaluate the performance

of the employed air quality model. The work carried out for both aspects of this project

is explained in detail and the utilised sources and tools are also described.

3.1 Monitoring methodology

This section incorporates the monitoring methodology adopted for all the required data.

The description of the site topography and the equipment that was used during the

measurement campaigns is followed by the data collection method regarding air

pollution and meteorological data, traffic flows and geometrical features at the

examined street canyon in Headingley. This information was gathered from various

sources and surveys, and was essential for the understanding of the local conditions as

well as for the second key part of the dissertation.

I.A. Karousos Modelling of Air Quality in Street Canyons 3. Study Methodology

- 31 -

3.1.1 Site and equipment description

The project concentrates on the study of a section of A660 Otley Road, which runs

through the centre of Headingley area in northern Leeds and comprises one of the

busiest radial routes towards the Leeds city centre (see also Figure 1.1). The traffic

conditions on this road show, therefore, high morning and afternoon peak periods as a

result of the numerous commuters using this route.

In this built-up area it is interesting to study the behaviour of air pollution, which is

increased due to the high traffic volumes on A660 and its dispersion is influenced to a

great degree by the surrounding topography. The segment of the road immediately

south of the junction with North Lane and Wood Lane is ideal for the investigation of

the street canyon effect on dispersion. Otley Road is a single carriageway in this area

with one lane per direction and the adjacent buildings are closely spaced and high

enough to create the conditions of a street canyon. The traffic lights at the major junction

are an additional factor affecting the vehicle operation mode (acceleration/deceleration)

and, thus, the exhaust

emission rates. An

aspect of the canyon

geometry and traffic

conditions during a

busy afternoon peak

hour is shown in

Figure 3.1.

Figure 3.1 View of the street canyon from the Otley Road /North Lane intersection.


- 32 -

The segment runs from Bennett Road to the signalised intersection for 85 metres and the

average height of the buildings is approximately 7m. It is a shallow canyon with height-

to-width ratio almost equal to 0.5 and its orientation is 328° with respect to the North.

There is one minor side-street (Shire Oak Street) in the middle of the east side of the

segment, which has very low traffic flows. There are, moreover, various shapes of

buildings with height exceptions and openings between them (see also Figure 3.2). The

position and the size of these obstacles, as well as those in the immediate

neighbourhood, have an impact on the pollution dispersion and will also be taken into

account in the modelling process.

Figure 3.2 Map of the immediate neighbourhood of the examined street canyon and of

the utilised monitoring points. (Source: Namdeo, A. (July 2006). Personal communication)

Street Canyon

(c) Crown copyright/database right 2005.An Ordnance Survey/EDINA supplied service


- 33 -

Figure 3.3 shows a typical cross-section of the street segment with average dimensions

as well as the pollutant measurement points. The street-level pollution was measured

by monitors mounted on the existing lampposts at the roadside, as part of a research

monitoring network set by Institute for Transport Studies, University of Leeds (see also

§3.1.2.1). Figure 3.2 contains the locations of some of those measurement devices (with

their ID numbers), whose data were also used in this project.

Figure 3.3 Vertical cross-section of the street canyon showing the monitoring points on

the lampposts.

The monitoring device used in this project is the LEARIAN Standard Streetbox. This

apparatus can provide real-time recording of local pollution levels. By use of

electrochemical sensors and a microprocessor controlled logger, it can monitor and

record information on various airborne pollutants, such as CO, NO2 and SO2. The logger

has the capability of recording up to several months of 15-minute averages; all data

collected are saved in its own memory and can be recovered from distance via its own

wireless communications system (Routesafe, 2006). The used device is illustrated in

Figure 3.4 and its detailed technical features are listed in Table 3.1.


- 34 -

.

Source: Routesafe, 2006

3.1.2 Data collection

The required measurements involve the collection of air quality, meteorological and

traffic data associated with the examined street and, by extension, the immediate urban

environment. The methods that were employed vary from manual surveys and use of

monitoring devices to data recovery from local and national databases.

Table 3.1 LEARIAN Streetbox characteristics.

Specification for Streetbox

Enclosure Weatherproof plastic 280mm x 168mm x 123mm deep

Weight 1.8 Kg Power Requirement 12V DC <50 mA Battery Life Six months

Sensors Up to three sensors: CO, NO2, SO2 Plus temperature & battery status

Memory 500 Kbytes as standard upgradeable to 8 Mbytes

Communications RS232 as standard; License exempt radio, Telephone Modem, GSM

Data format Full error checked self appended files

Inputs / outputs 8 analogue & 8 digital inputs in total 4 digital & counter outputs

Electrochemical Sensor Specification Gas Range Resolution Carbon monoxide 0-1000ppm 0.1ppm Nitrogen dioxide 0-10ppm 20ppb Sulphur dioxide 0-10ppm 25ppb Repeatability 1% of signal Span drift <10% per year Response time <40 seconds Temperature range -20 to +50ºC Sensor life up to 2yrs, depending on levels measured Specification for Base Station Ingress Protection IP64 Power Consumption 30mA; approx. 40 hrs on PP3

Figure 3.4 LEARIAN Streetbox No 133. Enclosure and example of fixing on a

lamppost.


- 35 -

3.1.2.1 Concentrations Data

Due to lack of time and monitoring equipment the roadside pollution monitoring was

limited to Carbon Monoxide measurements. Streetbox devices were attached to the

staggered street-light poles along the A660 route as well as at some off-road sites, at an

average height of 3.50m above the pavement. The street canyon geometry and the

position of some of the monitoring devices can also be seen in Figure 3.5. Ideally, if there

was another set of Streetboxes mounted onto the lampposts measuring emissions at a

lower level, the street canyon effect could be examined more meticulously. Moreover,

the distances of the lampposts from the roadside are not identical in all cases causing

thus some inconsistency in the measurements.

Figure 3.5 Looking north towards the junction and the locations of three Streetboxes.

The time coverage of the available roadside measurements was more than four months,

but the finally utilised data were from 1 May to 17 July 2006. This time period was

selected in order to minimise the missing data from the Streetboxes, but also because it

coincides with that of available meteorological data and, thus, comprises the simulation


- 36 -

period in the air quality modelling. The recordings are given at 15-minute intervals in

ppm (parts per million) and, so they were transformed into mg/m3 and hourly or even

daily averages in order to be comparable with the model outputs.

3.1.2.2 Meteorology and Background Data

Two essential datasets, which largely affect the pollution levels and behaviour at local-

scale dispersion phenomena, are the meteorological measurements and the background

pollutant concentrations. The set of meteorological data comes from the Leeds

Meteorological Mast in hourly averages as Microsoft Excel™ spreadsheets (Leeds City

Council (July 2006). Personal Communication). The meteorological station is fully

automatic and can record wind speed and direction, temperature, solar radiation and

relative humidity differing heights. It is located southeast of Leeds city centre away

from large obstacles that could bias local weather conditions and it has been approved

by the UK Meteorological Office.

The background pollution concentrations were obtained from various sources. Streetbox

No. 127 (see Fig. 3.2) provided urban background CO concentrations as it is located at

Wood Lane, which has negligible vehicle flows and is well away from the main roads.

The required background NOx, NO2 and O3 concentrations were downloaded from the

website of the UK National Air Quality Information Archive (AQ Archive, 2006).

3.1.2.3 Traffic Flows

Typical characteristics and volumes of traffic passing through the examined street

canyon are essential in order to be correlated with the temporal and spatial variations in

pollutant concentrations. The heavily congested peak periods, the presence of the

signalised junction at the north end of the canyon and the numerous bus lines operating

along A660 are the key factors to be considered in this air quality study.


- 37 -

There were two main sources of traffic data in this part of A660 road. University of

Leeds (UoL) and Leeds City Council (LCC) have provided diurnal vehicle flows south

of the signalised junction (Ropkins, K. (2006). Personal Communication). The flow data

were average values of manual counts and traffic loops data from four weekdays in

2004 and 2005, at half hour intervals and between 06:00 and 21:00. The residual diurnal

profile was completed by using the DfT database of traffic distribution by time of day

(Namdeo, A. (July 2006). Personal Communication) and factoring the missing data

(9pm-6am) with respect to the provided actual flows.

Manual classified traffic counts were, additionally, conducted at the same section during

AM and PM peak hours on 2nd and 3rd August 2006. The counts were carried out by two

stationary observers, one for each direction: northbound (from the city centre) and

southbound (towards the city centre). Appendix I contains the simple survey form,

where vehicles are categorised into five types: passenger cars, vans (LGVs), buses, rigid

and articulated lorries (HGVs), and the counts are divided into 15-minute intervals. The

up-to-date vehicle distribution could, thus, be obtained for the specific urban area of

Leeds and be used as input to the AQ models.

The counts were conducted south of the junction

in order to deduce the exact share of the

different vehicle types passing through the

canyon and emitting diverse pollutants at

varying rates. A manual survey involves,

however, a degree of human error, especially in

a busy traffic corridor. The use of two observers

and tally counters for all vehicle classes made

the counts easier and more accurate.

Figure 3.6 The utilised tally counter & the view from a survey point.


- 38 -

3.2 Modelling procedure

The computational part of the methodology involves the application of an air quality

model to analyse the pollution dispersion in the examined street canyon and predict the

temporal and spatial variations in CO concentration. The results of this procedure will

subsequently be compared with the monitoring outcome. The employed modelling tool

is described and, in particular, its configuration and the input data requirements, i.e.

traffic profiles, weather and background pollution data, site geometry and other

topographical parameters. The specific settings of the Windows based Operational

Street Pollution Model (WinOSPM), which was applied in this study, are

comprehensively reviewed below.

3.2.1 Parameterisation of WinOSPM

The first step of a new project in WinOSPM is the street configuration. The input data

for the examined street canyon in Headingley include:

the average height and width of the street canyon,

the distances from the location of the receptors (i.e. the monitoring points) to

either end of the street

the street’s orientation in relation to the north

the receptor height, and

the wind sectors with building height exceptions, i.e. the sections of the canyon

with differing height than the average one.

In Figure 3.7 the layout and the perspective of the canyon topographical features are

illustrated. Despite the existence of junctions and openings at both ends of the street

canyon, the buildings are set as if they continue lining up infinitely otherwise the model

will assume that there is an open area beyond these points.


- 39 -

(a) Street Geometry configuration window (b) Perspective view of canyon

Figure 3.7 Topographical features of the Otley Road street canyon.

The next task is the specification of the diurnal traffic flow distribution for all types of

vehicles. In this case, the input data consist of the traffic profile of the examined road

segment, which has been created by the combination of the diurnal traffic flows for all

vehicles (as provided by UoL & LCC data files) and the vehicle type distribution that

emerged from the manual counts in Headingley. The amount and the kind of the

observed traffic flows has prevented the formation of more realistic traffic profiles with

detailed distribution for every hour or separate cases for weekdays and weekends and

for different periods of the year. This shortcoming is actually the most important factor

affecting the accuracy of the model outcome and, thus, its evaluation against the

monitored emission levels.

The relevant table in the Traffic Data Editor (Figure 3.8) shows, apart from the hourly

distribution of the traffic for all vehicle types, the traffic speed for short and long

vehicles and the percentage of cold-starts. Due to lack of detailed information, these


- 40 -

inputs have been kept the same as the default ones for streets passing through mixed

areas (Type D of the default Danish traffic files). The diagram of average diurnal traffic

volumes (in vehicles per hour) for passenger cars, vans and all vehicle types is

illustrated in Figure 3.9.

Figure 3.8 Traffic Data Editor window for the hourly traffic distribution in Otley Road.

Figure 3.9 contains the average diurnal CO emissions (in grams per km per hour)

corresponding directly to the traffic profile. The calculation of traffic emissions in

WinOSPM is based on the traffic flow (in vehicles/hour) and the emission factors (in

g/vehicle/km) for each pollutant and vehicle type. As previously described, emission

factors (EF) can differ considerably for the various vehicle sub-categories (depending on

engine size and technology, fuel type, mileage etc.) and the method that is incorporated

in WinOSPM for calculation of these factors is based on the European Emission Model

COPERT-III (see also Ntziachristos and Samaras, 2000).


- 41 -

(a) Flow (veh/h) (b) CO (g/km/h)

Figure 3.9 (a) Average diurnal traffic flows & (b) CO emissions for three vehicle categories. (adapted from WinOSPM Traffic Editor window)

The categorisation of vehicles is based on the national fleet share for each country and

for this application, these data have been adopted from the UK National Atmospheric

Emissions Inventory for the year 2006 (Namdeo, A. (July 2006). Personal

Communication). The basic emission factors are functions of the average traffic speed in

the street, which in our case is assumed to be 50km/h, i.e. approximately equal to the

30mph speed limit on Otley Road. The factors that are involved in the local emission

estimates can emerge after the correction expressions have been applied, relevant to the

effect of cold-starts, increasing vehicle mileage and fuel composition. (Gokhale et al.,

2005). These expressions have been kept identical to the default data of WinOSPM (i.e.

for Danish conditions), since they are adopted in European level and can be used for UK

conditions as well. In this study only the cold-start corrections have been taken into

account. The operational degradation of the catalytic converters with the mileage of the

vehicles and the fuel composition corrections for CO emissions have been ignored.

In Figure 3.10 an illustrative example of vehicle EF expressions is given for UK and for

year 2006, as it can be configured in WinOSPM. In the upper section the national fleet

share information is shown for vehicle and fuel types, modified for UK conditions. The

lower part contains all the relevant parameters that can be applied to estimate emission

-A- All Vehicles -B- Passenger Cars -C- Vans


- 42 -

factors; the case of

the EF function for

carbon monoxide is

here illustrated for

EURO I petrol cars

with engine size

smaller than 1.4

litres.

Figure 3.10 WinOSPM configuration window for Vehicle Emission Factors.

The second key category of WinOSPM input data involves meteorological and

background pollution data. A user provided data file containing variables, such as wind

speed and direction, the prevailing temperature and the global radiation levels, has been

created from the information collected from Leeds Met Station. The time coverage of

these data was about 2.5 months, defining thus the duration of the simulation period.

The same file includes general background pollution concentrations in the urban area,

which were downloaded from the monitoring equipment and the online resources

regarding the area of Headingley. Figure 3.11 illustrates the list of variables included in

the user-defined WinOSPM input file.

Figure 3.11 Variables list in the weather and background input file of WinOSPM.


- 43 -

The ideal weather and background data would be measured from a building roof along

the street canyon. The available data are, however, assumed to represent adequately the

conditions above roof level in the city. Additional limitations are attributable to possible

missing hourly data in the meteorology and urban background file, as the calculations

for the particular hours will be skipped.

3.3 Data analysis

Regarding the data analysis methodology, the primary aim is to acquire results

explanatory for the specific conditions of the study and in consistent, comparable

formats with each other. The conducted statistical analysis of the air quality monitored

and modelled data is contained in the next chapter including the outputs from Microsoft

Excel™ software, which was broadly utilised in order to process and graphically

visualise the results.

The analysis starts with the examination of the observed traffic and meteorological data.

Subsequently, the pollution data from the Streetboxes are processed, in order to depict

the actual temporal and spatial variation of CO concentrations within the street canyon.

Hourly and daily average pollution levels are compared between monitoring points

along and across the street and the presence of the canyon effect is discussed.

The processing of the model outputs involves descriptive and regression statistical

analysis of the predicted data, comparison between monitored and predicted

concentrations and model validation techniques.

Chapter 4

Monitoring and Modelling Results

The possible factors affecting pollutants emission and dispersion in urban street canyon

conditions have been previously described. In this chapter the analysis of the data, that

were either collected or calculated during this study, attempts to explain the

relationships between those factors and the street-level CO concentrations. The traffic

and meteorological data are key elements of the analysis as their dynamics influence

pollution levels to a great extent in densely populated and structured urban areas, such

as Headingley.

4.1 Traffic and meteorology

4.1.1 Traffic Flow

The available data on traffic volumes for both directions of Otley Road cover significant

parts of four weekdays in 2004 and 2005 and have been used to produce one average

diurnal traffic flow profile for the period between 6:00am and 9:00pm (orange area in

Figure 4.1). The AM peak period around 08:00 is obvious in the diagram and is typical

for a radial route towards the city centre. The traffic volumes more or less stabilise

during the rest of the day and a less sharp PM peak occurs between 18:00 and 19:00 in

the evening again due to the high commuter flows originated from the city centre and

passing through Headingley area.

I.A. Karousos Modelling of Air Quality in Street Canyons 4.Monitoring & Modelling Results

- 45 -

0

500

1000

1500

2000

06:00 07:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00

Time of Day

Traf

fic F

low

(veh

icle

s/hr

)

Figure 4.1 Observed fractional traffic counts and average diurnal profile on Otley Road.

(adapted from Ropkins, K. (July 2006). Personal Communication)

The hourly traffic data for the remaining hours of the day are required as inputs for the

air quality models in order to factorise emissions over a 24h period and to achieve

complete and accurate results. The method described in section 3.1.2.3, involves the use

of factors generated by the UK average traffic distribution by time of day on all roads

and tailored to the above observed flows. The results are listed in Table 4.1, where the

sum of the observed flows has been divided by the sum of the corresponding hourly

ratios of AADT for a typical weekday to provide the factored-up AADT (Annual

Average Daily Traffic) for the whole day (approximately 35,884 vehicles per day). This

figure is then distributed to each hour in order to fill-in the missing data from the

diurnal profile of Figure 4.1.

Figure 4.2 illustrates the finalised average diurnal traffic profile created, which has been

used in the dissertation for investigation of the pollution behaviour and correlation to

traffic fluctuations, but also as input user-defined data into the street canyon model. It is

marked that during the late evening and early morning hours (9pm-5am), the traffic

flows are very low, although in the weekend the evening traffic is expected to be higher

due to the presence of many pubs and bars in the immediate neighbourhood.


- 46 -

Table 4.1 Original, factored-up and completed hourly traffic flows.

Time Observed Flow

Ratio of typical AADT

Factored-up diurnal profile Completed data

00:00 - 01:00 - 0.007 238.476 238.48 01:00 - 02:00 - 0.005 168.336 168.34 02:00 - 03:00 - 0.004 140.280 140.28 03:00 - 04:00 - 0.004 154.308 154.31 04:00 - 05:00 - 0.007 238.476 238.48 05:00 - 06:00 - 0.015 533.065 533.06 06:00 - 07:00 890.00 0.036 1290.578 890.00 07:00 - 08:00 2609.00 0.068 2454.903 2609.00 08:00 - 09:00 2874.87 0.075 2693.380 2874.87 09:00 - 10:00 2438.53 0.062 2216.427 2438.53 10:00 - 11:00 2221.80 0.057 2062.119 2221.80 11:00 - 12:00 2183.93 0.058 2076.147 2183.93 12:00 - 13:00 2180.85 0.059 2118.231 2180.85 13:00 - 14:00 2207.15 0.061 2188.371 2207.15 14:00 - 15:00 2195.30 0.063 2272.539 2195.30 15:00 - 16:00 2133.05 0.068 2454.903 2133.05 16:00 - 17:00 2224.80 0.077 2763.520 2224.80 17:00 - 18:00 2093.67 0.079 2847.688 2093.67 18:00 - 19:00 2285.00 0.065 2314.623 2285.00 19:00 - 20:00 2186.00 0.045 1627.250 2186.00 20:00 - 21:00 1793.00 0.032 1136.270 1793.00 21:00 - 22:00 - 0.023 841.681 841.68 22:00 - 23:00 - 0.018 631.261 631.26 23:00 - 24:00 - 0.012 420.841 420.84

Total (06:00-21:00) ΣΟ = 32516.95 ΣR = 0.906

Total (24-hour) - 1.000 ΣΟ/ΣR = 35883.67 AADT = 35883.67

0

500

1000

1500

2000

2500

3000

3500

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00Time of Day

Traf

fic F

low

(veh

icle

s/hr

)

Completed profile for Otley RoadTypical UK profile

Figure 4.2 Completed diurnal flows and associated typical profile for AADT ≈ 35,884.


- 47 -

4.1.2 Vehicle Type Distribution

The data collected from the conducted manual classified counts are analysed in order to

generate the distribution of different vehicle types that is typical for the specific street

segment on Otley Road, south of the junction with North and Wood Lane. Although the

sample is of limited size it is representative of the traffic conditions in the area. Figure

4.3 contains: (a) the up-to-date modal split as formed by the current survey data, (b) the

respective modal split at the same location estimated two years ago (Ropkins, K (July

2006) Personal Communication), and (c) the Urban Vehicle Estimated Distribution for

UK in 2006, as adopted from National Atmospheric Emissions Inventory (NAEI, 2006).

(a)

Buses, 3.4%

Artic HGVs, 0.7%Rigid HGVs, 2.2%

LGVs, 6.6%

Cars, 87.1%

(b)

Car/Van, 92.5%

Motorcycle, 1.0%Lorry, 2.3%

Bus, 4.2%

(c)

Buses, 1.5%

Artic HGVs, 0.6%

Rigid HGVs, 2.2%

LGVs, 10.6%

Cars, 85.1%

Figure 4.3 Estimated vehicle distribution in Headingley for (a) 2006 and (b) 2004, and (c)

for current UK national urban conditions.


- 48 -

It is evident from the above distributions that passenger cars are the vast majority of

vehicles along the street canyon (87.1%) and, hence, comprise the predominant source of

exhaust emissions and also mechanical turbulence in the confined urban environment.

The general pattern remains similar to that of 2004 data and the main differences, except

for the separate modal shares for passenger cars and vans, are the slightly lower

proportion of buses (-0.8%), presumably due to less frequent bus services during the

period of the 2006 survey. Nevertheless, the bus share is more than two times that of the

UK national database, which is characteristic of the examined radial road and will have

an effect on the rates and types of emitted pollutants in the area.

4.1.3 Meteorological data

A wide range of meteorological parameters were provided from Leeds City Council Met

Station and the time coverage was from 1st May to 17th July 2006 with hourly resolution.

The information is assumed to correspond to the average conditions in the examined

urban area. For instance, Figure 4.4 contains the frequency distribution of the provided

wind speed measurements. The average wind speed is 2.94m/s, whereas the percentage

of low wind speeds (defined as u ≤ 2 m/s) is 32.7% and calm conditions (u ≤ 1 m/s)

account for 7.8% of the observations.

The wind rose in Figure 4.5 shows that during the examined 2 ½ months period the

predominant wind direction is West-Southwest followed by quite frequent Eastern

winds. These prevailing wind flows are perpendicular (or near-perpendicular) to the

street canyon axis, which means that vortex formation and uneven CO concentrations

across the canyon should be expected for the period studied.


- 49 -

0

0

3

1.5

6

3.1

10

5.1

16

8.2

(knots)

(m/s)

Wind speed

80

0

0.01

0.02

0.03

0.04

0.05

0.06

0 1 2 3 4 5 6 7 8 9Wind Speed (m/s)

Freq

uenc

y D

istri

butio

n

Figure 4.4 The frequency distribution of the wind speed measurements.

0° 10°20°

30°

40°

50°

60°

70°

80°

90°

100°

110°

120°

130°

140°

150°160°

170°180°190°200°

210°

220°

230°

240°

250°

260°

270°

280°

290°

300°

310°

320°

330°340°

350°

50

100

150

200

Figure 4.5 Wind rose for the examined time period.

4.2 Streetboxes data analysis

The monitoring campaign of CO concentrations across the area of Headingley has been

running since February 2006, but the data that were downloaded and analysed in this

1 May - 17 July, 2006


- 50 -

study cover approximately two and a half months of continuous measurements at 15-

minute intervals. As mentioned earlier, this period corresponds to the available

meteorological data and, hence, the model input and output data.

The monitoring devices are located along the A660 and the ones used in this study have

the ID numbers 134, 125, 133 and 146, whereas Streetbox No. 127 is located at a

background site away from busy arterial roads (as shown in Fig. 4.6). The downloaded

raw data of CO concentrations (in ppm) are given in 15-minute intervals and were

transformed into hourly averages for practical reasons. Figure 4.6, however, shows the

overall mean concentration measured from each Streetbox. It is evident that pollution

levels are increasing as we are approaching the congested signalised intersection and

the more confined urban environment of the street canyon.

The observed concentrations of carbon monoxide should be strongly correlated with the

relevant traffic volume as it is a primary pollutant. The average diurnal CO levels

shown in Figure 4.7 appear to agree generally with the traffic profile. The sharp increase

due to the morning peak period is clearly illustrated, whereas some of the Streetboxes

(No. 134 & 133) demonstrate even higher afternoon concentrations. These monitors are

all located at the eastern (right-hand) side of the street and the rest (No. 125 & 146) at the

western footway. A most probable reason for this incident is the presence of the traffic

lights at the main junction. As it was also observed during the manual surveys, in the

afternoon the outbound stream usually gets congested and, hence, the exhaust

emissions from the stopping/starting vehicles increase drastically. The western

Streetboxes (SB 125 & 146) appear to be directly affected by the queue before the

junction and, hence, show constantly high concentrations during the whole day. The

devices on the other side mainly monitor emissions from almost free flow traffic on the

southbound direction, so the PM congestion on the other lane affects considerably the

measured pollution levels. These are, however, influenced by other factors, such as

prevailing wind speed and direction, at the same time period.


- 51 -

*urban background pollution level

Figure 4.6 Average measured CO concentration by the installed Streetboxes.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00

Time of Day

CO

(ppm

)

0

1000

2000

3000

4000

5000

6000

7000

Vehicles/hour

Flow

SB 134

SB 125

SB 133

SB 146

Figure 4.7 Average diurnal CO measurements at each Streetbox vs. traffic flow profile.

(c) Crown copyright/database right 2005.An Ordnance Survey/EDINA supplied service

134 125 133 146127*

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

CO

(ppm

)

Streetbox


- 52 -

Regarding the atmospheric flows, which play crucial part in the pollutants dispersion

also at street-level, the wind speed and direction are the relevant parameters that can be

investigated. The wind sectors that are of main interest are those including winds

perpendicular to the street axis, so that the canyon effect can be examined. The wind

direction dependence of measured CO concentrations has been investigated and the

pollution levels appear to be little influenced by the wind flow (see scatter diagrams in

Appendix II). According to the street canyon wind vortex theory, the dispersion should

reduce the CO concentrations for the windward side of the canyon. This dependence is

fairly pronounced for SB 134 and 133, but it appears not to be the major dispersion

mechanism in this case.

Nevertheless, when only the CO measurements for near-perpendicular (90±22.5°) wind

directions are averaged for the examined time period, the street canyon effect is more

evident. Figure 4.8 shows the mean CO concentrations for the two respective wind

sectors (NE and SW) and

clearly demonstrates that in

all cases the pollution level is

increased for the leeward

monitoring point.

The investigation of carbon

monoxide levels against the

wind speed data (Appendix

II) shows that the expected

inversely proportionate

relationship is not palpable for these measurements. The most representative data come

from the diagrams of SB 125 and 146, which display that higher CO levels, to an extent,

occur during lower wind speed cases.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

SB 134 SB 125 SB 133 SB 146

CO

(mg/

m3)

Figure 4.8 Mean CO values for the examined winds sector.


- 53 -

4.3 Monitored versus modelled results

The WinOSPM street canyon model generally utilises hourly time series of street-level

emissions, urban background pollution data and meteorological data as input values,

and produces predictions of concentrations at the predefined receptor locations, for the

corresponding simulation period and the selected scenario parameters. The original

model was applied with user-provided traffic and meteorological data and without any

modifications in the incorporated emission model, except for the updated UK national

vehicle fleet share for the considered year of 2006. No adjustments or calibration of the

model were employed according to the measured data from this project. Table 4.2 is

adapted from the calculation window of WinOSPM and contains the summary of the

input and output data after the model is run.

The main sources of disparities between observed and predicted values are the

inaccuracies in the exhaust emission calculations and the dispersion modelling results

due to the generalised values used for many of the nationwide and local-scale model

parameters. On the other side, accuracy is decreased by a wide range of error types

involved in the methodology of traffic, meteorology and pollution data collection and,

naturally, by the random mechanism of atmospheric turbulence (Kukkonen et al., 2001).

Table 4.2 WinOSPM Results Summary Table.

Street: OtleyRd Calculated on: 02/08/2006 17:06:43 Average Daily Traffic: 35884 (Calculated); User provided [C:\Program Files\WinOSPM\Data\Traffic\National\ UK\MyTypeD2.txt] Emission Scenario Year: 2006 Period Covered (User provided Meteorological Data): 01. May 2006 00:00 - 17. July 2006 23:00 Urban Background: User provided

Component: Hourly Max Daily 8 hours mean Daily Averages

CO (mg/m³) Mean Max 98%-ile

99.8%-ile Max 93.2%-

ile Max 90.4%-ile

98%-ile

Data Coverage

(% of year)

Street Modelled 0.84 1.54 1.21 1.41 1.24 1.16 1.06 0.98 1.04 21.37 Background 0.74 1.33 1 1.24 1.11 1.02 0.97 0.87 0.94 21.37 EU Limit Value(2005) 10 75


- 54 -

The time resolution of the modelled CO concentrations is hourly. The averaged diurnal

CO predictions for the whole simulation period are illustrated in Figure 4.9 along with

the input traffic data. The Receptor 1, which is located at the eastern side of the canyon,

displays constantly lower CO concentrations compared with Receptor 2 across the

street; an initial sign of the street canyon effect. Taken into account the prevailing wind

direction (see Fig. 4.6), Receptor 1 (R1) is more often on the windward side of the canyon

receiving, thus, flows of fresh air which reduce the average pollution level on that side.

The association of this diagram with the one in Fig. 4.7 reveals that the model results fit

better to the measurements from the western monitoring points. The measured values

across the street are affected by the PM congested period at the upstream signalised

junction. This effect cannot, though, be represented in WinOSPM.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

00:0

0

01:0

0

02:0

0

03:0

0

04:0

0

05:0

0

06:0

0

07:0

0

08:0

0

09:0

0

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

Time of Day

CO

(mg/

m3)

0

1000

2000

3000

4000

5000

Vehicles/hour

Flow

Receptor 1

Receptor 2

Figure 4.9 Average diurnal CO predictions at each receptor point vs. traffic flow profile.


- 55 -

The observed and modelled pollutants behaviour is demonstrated in the following

graph of hourly average CO concentrations for seven consecutive days of one random

week in May 2006 (Figure 4.10). It contains the predicted CO values for the two opposite

monitoring points (R1 & R2), which are also compared with the average observed values

from the four Streetboxes on Otley Road. It is evident that the modelling time-series

results follow the same pattern as the averaged in the previous graph. R1 concentrations

are lower than R2 for the largest part of the day, but overall during night-time these are

almost equal, because of the low traffic volume and possibly the lower wind speed

cases. The comparison with the monitoring results reveals that the model, generally,

displays the same diurnal behaviour as the real pollution levels, but represents poorly

the partial peak intervals. In other words, it underestimates CO concentrations during

day-time and overestimates them in the evening and night hours. The potential reasons

behind this are the generalising assumptions about the traffic flow and vehicle

distribution on A660, as well as the meteorological data, which cannot entirely represent

the above-roof and street-level conditions in this urban area.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21

15/05/06 16/05/06 17/05/06 18/05/06 19/05/06 20/05/06 21/05/06

Days of Week 15-21/05/06

CO

(mg/

m3)

Receptor 1

Receptor 2

Mean Observed

Figure 4.10 Observed and predicted hourly CO concentrations from

Monday, 15th until Sunday, 21st May 2006.


- 56 -

The evaluation of the modelling results against the CO measurements is also conducted

via the scatter diagrams between corresponding monitoring points. For example,

Receptors 1 and 2 are compared with Streetboxes 134 and 125 respectively, as these are

the closest points with available observed and predicted concentrations. Figure 4.11

shows these plots of hourly CO concentrations at street level. Table 4.3 contains the

results of the statistical analysis between the observed and predicted CO values. The

ratios of measured mean concentrations to modelled ones are 1.248 for R1 and 1.154 for

R2. This is an initial proof that model generally underpredicts the average values for the

examined period. The correlation coefficient values R and R2 for the scatter plots are

very low, showing hence the weak correlation of the predicted and observed time series

of concentrations at the examined locations.

For model performance evaluation it is, however, suggested to use additionally an index

of agreement (d), which determines the level of correlation between magnitudes (and

signs) of the monitored values about observed mean and the predicted deviations about

observed mean. The equation for d is:

[ ]21872

1

1872

1

2

1

∑

∑

=

=

−+−

−−=

i

i

OOiOPi

OiPid

)()(

)( and 0 ≤ d ≤ 1,

where Pi is the ith predicted value, Oi is the ith observed value and O the mean of

observed values (Gokhale et al., 2005).

The outcomes show that the index of agreement is slightly better for the R1 – SB134 than

for the R2 – SB125 comparison, which is the opposite of what the R2 coefficient suggests.

The combined results of the statistical analyses conducted indicate that the modelled

concentration levels are not in good agreement with the measured ones and show a high

potential for error in the procedure. The limitations of the monitoring campaign and the

simplification of some types of input data consist the main sources of these inaccuracies.


- 57 -

R1 vs. SB134

y = 0.1098x + 0.7005R2 = 0.1532

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

Measured (mg/m3)

Mod

elle

d (m

g/m

3)

R2 vs. SB125

y = 0.1785x + 0.6706R2 = 0.4715

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

Measured (mg/m3)

Mod

elle

d (m

g/m

3)

Figure 4.11 Scatter plots of measured against predicted CO values for the two pairs of monitoring points.

Table 4.3 Results of statistical analyses between observed and predicted CO values.

Indicators R1 vs. SB134 R2 vs. SB125 Ratio of mean concentrations 1.248 1.154 Correlation Coefficient R 0.391 0.687 Square Correlation Coefficient R2 0.153 0.471 Index of agreement d 0.702 0.518

Meteorological analysis

The dispersion of air pollution in a street canyon is highly correlated to the local

atmospheric conditions, which determine whether a stable wind vortex is formed

between the canyon boundaries. Under such circumstances a substantial concentration

gradient across the street might be observed, with the windward side usually displaying

lower pollution levels than the leeward side. The evaluation of the model performance

requires the analysis of the behaviour of the modelled pollution levels when compared

with the input meteorological data. In our case, the computed CO concentrations are

compared against the hourly data of wind speed and direction available for the

simulation period.


- 58 -

The following table contains a summary of the measured and modelled mean

concentrations classified with respect to wind speed, i.e. for u≤2m/s, 2<u≤4m/s and

u>4m/s. The predicted values show that, in low wind speed cases, the pollutants

dispersion is poorer resulting generally in higher street-level pollution concentrations.

This is not, however, evident in the observed concentrations, as indicated earlier.

In Appendix II, the scatter plots illustrate the inverse relationship between the predicted

CO values at both receptors and the wind speed, which is though not pronounced. The

scatter diagrams of the modelled data against the wind direction data (also included in

Appendix II) also illustrate a weak dependence on the varying direction of the above-

roof wind flows.

Table 4.4 Wind speed dependence of the observed and predicted average CO levels.

Average CO Concentration (in mg/m3)

Predicted Observed Wind Speed R1 R2 SB134 SB125 SB133 SB146

u ≤ 2m/s 0.841 0.863 0.879 0.845 0.724 0.762

2 < u ≤ 4m/s 0.818 0.849 0.997 0.958 0.755 0.740

u > 4m/s 0.789 0.828 1.071 1.034 0.758 0.748

In Figure 4.12 a combination of the wind speed and direction data attempts to show

their correlation with the modelled concentrations at the two receptor points of

WinOSPM. The wind direction data have been classified into two sectors, where the

wind flow is perpendicular (90°) or near-perpendicular (±22.5°) to the street axis (as

shown in the canyon layout, right insert). Only these data were consequently separated

into two speed categories (u ≤ 2m/s & >2m/s) and the average values for each


- 59 -

subcategory and each receptor are illustrated in the bar chart. The differences between

the windward and the leeward side of the street are quite small, indicating thus that

wind flows were not the main dispersion mechanism in this case. Only for SW winds a

weak recirculation vortex appears to transfer the pollution to the leeward side (R2) as a

result of the street canyon effect. The small aspect ratio of the canyon (H/W≈0.5) and the

assumptions made for the available meteorological data are the main reasons for the

poorly depicted wind flow dispersion phenomena within the examined urban area.

0.00

0.20

0.40

0.60

0.80

1.00

R1 R2

CO

(mg/

m3)

Figure 4.12 The dependence of the average CO predictions on combined wind speed

and direction clusters. The shaded bars indicate the leeward side in each case.

4.4 Summary

There is a general endeavour at European level to produce and assess air quality

modelling results from case studies in order to actually utilise the evaluated models for

the various active air quality strategic plans. The results from applications of WinOSPM

model in a wide range of urban environments and research programmes have showed

that it is capable of being used as a successful modelling tool as its predictions are often


- 60 -

in good agreement with those from the corresponding monitoring campaigns (Gokhale

et al., 2005). The model’s performance is, however, highly dependent on the local

conditions and the availability of up-to-date and explicit input data. In this case, the

results emerging from the monitoring procedure suggest the presence of mild canyon

effect, but the modelling outcomes demonstrate that this phenomenon is not the chief

dispersion mechanism within the examined road segment. The WinOSPM appears,

moreover, to underpredict the CO concentrations when compared to the results from

the ongoing monitoring campaign in the area of Headingley. The potential factors

affecting the model performance and the relevant evidence from the conducted analysis

are further discussed in the next chapter.

Chapter 5

Discussion and Recommendations 5.1 Synopsis

Despite the recent technological improvements on vehicle emission control and the

increasingly stricter regulations, air pollution produced by road transport is one of

the most pressing problems in modern urban agglomerations both in the developed

and the developing world. Air quality limit values, whose objective is to protect

public health, are frequently exceeded, particularly in busy urban streets and other

hotspots in built-up areas. Evidence is continuously emerging related to the high risk

of human exposure to increased pollutant concentrations in densely populated urban

areas in contrast with the proven adverse effects on human health. It is, hence,

imperative to fully understand the pollutants behaviour within confined urban

surroundings in order to achieve further emissions abatement and improvements in

urban air quality.

This study presented the results of a limited monitoring and modelling

methodology, which was adopted in order to understand the predominant

mechanisms of pollution dispersion in an urban street canyon. The examined road

segment of A660 in Headingley is relatively narrow and flanked by closely spaced

buildings on both sides. Evidence of the street canyon effect on the diffusion of

Carbon Monoxide emissions was, hence, attempted to be established via the analysis

of the observed concentrations and the results of the application of OSPM dispersion

model. The relationship between the engine emission rates, the ambient conditions

I.A.Karousos Modelling of Air Quality in Street Canyons 5.Discussion & Recommendations

- 62 -

and the other factors affecting the pollutants dispersion could not be spherically

examined because of the shortcomings in both the monitoring and the modelling

procedure. Moreover, the model predictive performance of CO levels was proven to

be poor, when evaluated against the observed concentrations. WinOSPM generally

underpredicted the pollution measurements and its sensitivity to wind speed and

direction changes could not reveal clear evidence of the street canyon effect. The

limitations and the findings of the performed analyses are criticised in this chapter.

5.2 Conclusions and discussion

This project was divided into two key sections: the experimental part, involving the

monitoring campaign in the urban area of Headingley, and the computational

methodology, which consists of the application of the WinOSPM street canyon

model and the relevant analysis of the results.

Regarding the first part, the post-processing of the observed data has shown that the

average traffic profile for Otley Road in this area is characteristic of a congested

urban location with morning and evening peak periods. The vehicle type distribution

was also representative for this road due to the up-to-date manual classified counts

south of the main signalised junction. The supplied meteorological data covered a

fairly long period of time and the installed LEARIAN Streetboxes provided the

corresponding roadside CO concentrations from various points in the examined area.

Nevertheless, the limitations of the monitoring campaign were considerable with

regard to the representation of the real conditions in the street canyon and,

consequently, the evaluation of the modelling performance. The diurnal traffic

distribution was complemented for the evening and night hours (from 9pm to 6am)

by use of the national typical hourly profile for urban streets. This adjustment and

the relatively small sample of the provided traffic data cause deficiencies in the

correlation between the actual vehicle flows and the traffic-related pollution levels.


- 63 -

Moreover, the deduced from the manual counts modal split in the area corresponded

accurately to the AM and PM surveying periods, but not to the 24-hour profile which

was used as input to WinOSPM model.

The assumptions made for the meteorological data were an additional source of

inconsistencies to the study. The Leeds Met Mast measurements included wind

speed and direction data, which theoretically represented the above roof level wind

flow characteristics in Headingley, whereas the temperature and global radiation

data would correspond to the street-level conditions of the canyon. The urban

background pollution should normally be measured at roof level too, but due to lack

of resources the Streetbox 127 located at a low trafficked side street provided the

necessary background levels.

As demonstrated in the comparative analysis of monitored versus modelled results, a

plethora of additional data could have been used as inputs to WinOSPM to increase

the accuracy of the methodology. The largest part of the emissions modelling

remained unchanged according to UK and EU typical standards. For instance, hourly

data of vehicle type distribution, average speeds for short and long vehicles and cold

start percentages as well as vehicles average mileage information were not available

and, thus, the default values of WinOSPM were taken into account for the emissions

estimates. Furthermore, the congestion observed at the northbound lane of A660

because of the signalised intersection with North Lane has a major impact on the

gaseous emission rates along this section, which could not be represented in the

present application. Information on proportions of average idling, acceleration/

deceleration and cruising time, would amend the modelled vehicle emissions.

The dispersion modelling is also affected by a variety of parameters. The lack of

accurate data related to street and roof level atmospheric conditions, the above

mentioned shortcomings in the emissions calculation and the background pollution

data are believed to influence mostly the model performance. The small aspect ratio

of the street canyon, which may even prevent the formation of a primary wind


- 64 -

vortex when the wind is perpendicular, the lack of knowledge related to the vehicle

induced mixing and the thermal turbulence along the road are additional factors,

which might predominantly affect the dispersion mechanism in the specific canyon,

but it is not possible to be parameterised in the model.

The outlined limitations had a negative impact on the model performance against the

monitored values. The underestimates of CO concentrations are, however, mainly

caused by the inaccurate traffic data, since carbon monoxide emissions are primarily

related to the traffic volumes and their diurnal fluctuations. Secondly, the

meteorological and background data play important role in the final outputs of the

street canyon model. Finally, experimental results have shown that unusual airflow

and dispersion mechanisms (e.g. meandering, counter-rotating or secondary

vortices) occur near the street outlets, such as the two ends of the canyon or the gaps

between buildings, and affect the pollutant concentrations (Tsai and Chen, 2004).

The fluidyn-PANACHE model

The initial scope of this study included the application of a second air quality model

for the street canyon in Headingley in order to assess its predictive performance

against the monitored CO levels, but also against the results of OSPM. The

evaluation of the two models could then demonstrate which one fits better to the

local conditions and morphology of the examined street canyon.

This second model is Version 3.3.2 of fluidyn-PANACHE-PANAIR developed by

TRANSOFT international. Fluidyn-PANACHE is a self-contained software package

designed especially for simulation of atmospheric pollution. Using well-established

fluid dynamics principles, PANACHE predicts the spread of different types of

pollutants released to the atmosphere under different conditions. Pollutant

dispersion is influenced by the prevailing terrain and weather conditions. The user

can supply all the relevant input data and conduct the simulation. Results of the

simulation can be viewed in graphic details using the post-processing facilities

provided.


- 65 -

Nevertheless, due to technical problems of the simulation routine within the model it

was impossible to generate any reliable results of predicted values inside the

computational area. Part of the setup of the model terrain with the user-defined road

and urban area sources of pollutants are shown in Appendix III. This endeavour

could, however, be continued in future work (see also next section).

5.3 Recommendations for further work Based on the weaknesses identified during the assessment of the modelling results,

the key recommendation for future work would certainly involve a more detailed

local-scale monitoring campaign. Comprehensive collection of traffic data with

vehicle type and engine classification, by extensive use of Automatic Number Plate

Recognition (ANPR) and/or traffic loops, would provide an accurate database for

the calculation of emissions, minimising hence the potential error. The adjustment of

the emission factor functions with corrections for cold-start percentages and vehicle

mileage would also reduce the inaccuracies of the emission modelling.

The comprehensive measurement of the required street and roof level meteorological

and urban background data would provide the associated time series input data for

the dispersion modelling process. Most of these observations will be readily available

through the Instrumented Junction project run by the Institute for Transport Studies,

Leeds University. The signalised junction of Otley Road with North Lane and Wood

Lane will be soon fully equipped with monitoring devices of the main air pollutants,

meteorological instruments, ANPR cameras and traffic loops, which will allow for a

highly detailed and spherical monitoring campaign in the congested urban area.

On the modelling side, helpful and possibly more accurate results would emerge

through the use of the multi-street feature offered in WinOSPM, although this has

not been widely used and assessed. The inclusion of neighbouring main and side

streets in the dispersion model may give more representative outputs, especially

when compared with monitoring points near the street outlets. A more complete


- 66 -

picture of the air quality in the examined street canyon would result by modelling in

OSPM the rest of the main air pollutants, i.e. nitrogen oxides (NOx, NO and NO2),

ozone O3, particulate matters PM10 and benzene. The relevant observed levels should

obviously be available to make possible the model performance evaluation. Finally,

the application of other air quality models of various types (empirical, computational

etc.) could provide data for comparative analysis and evaluation of the employed

modelling tools in order to find the optimal one for the local conditions.

- 67 -

Acknowledgements

Firstly, I would like to sincerely thank my supervisor, Dr Anil Namdeo, for his

support and advice from the start of this study until the full completion. His helpful

guidance and all the information he offered made this project possible.

Secondly, I am indebted to the people who provided me with all the necessary data

that were used in this project, i.e. to Karl Ropkins (ITS Research Fellow), Leonardo

Sabatino (ex Marie Curie student from University of Palermo) and David Young

(PhD Student, Leeds University) for the traffic survey data at the Instrumented

Junction in Headingley, as well as to Leeds City Council and, especially to Richard

Crowther, for the traffic loop records and the meteorological and background data I

have been provided with.

I would also like to thank Colin Oates for downloading the Streetboxes data and

helping me to prepare for the traffic survey. Finally, special thanks are due to my

colleague and good friend Georgios Anastasiou for his help and for taking part in the

traffic counts on Otley Road.

- 68 -

References

AQ Archive (2006). The UK National Air Quality Information Archive [online]. Available at:

http://www.airquality.co.uk/archive/index.php [Accessed on 12th July 2006]

Berkowicz, R., Palmgren, F., Hertel, O. and Vignati, E., (1996). Using measurements of air

pollution in streets for evaluation of urban air quality—meteorological analysis and model

calculations. The Science of the Total Environment 189/190, pp. 259–265

DEFRA (Department for Environment, Food and Rural Affairs) (2003). The Air Quality

Strategy for England, Scotland, Wales and Northern Ireland: Addendum. Published by the DEFRA

in partnership with the Scottish Executive, The Welsh Assembly Government and the

Department of the Environment in N. Ireland

DEFRA (2006). Source publication: e-Digest of Environmental Statistics, Air Quality [online].

Available at: http://www.defra.gov.uk/environment/statistics/index.htm [Accessed on 16th

June 2006]

EC-MEET (1999). MEET - Methodology for calculating transport emissions and energy

consumption. Transport Research – 4th Framework Programme. European Commission,

Luxembourg: Office for Official Publications of the European Communities

EEA (European Environment Agency) (2006a). Environmental themes: Transport [online].

Available at: http://www.eea.europa.eu/main_html [Accessed on 20th July 2006]

EEA (2006b). Air pollution at street level in European cities. European Environment Agency,

Technical Report No1/2006, ISSN 1725-2237

EEA (2006c). Transport and environment: facing a dilemma. TERM 2005: indicators tracking

transport and environment in the European Union. EEA Report No 3/2006, ISSN 1725-9177

Gokhale, S.B., Rebours, A. and Pavageau, M. (2005). The performance evaluation of WinOSPM

model for urban street canyons of Nantes in France. Environmental Monitoring and Assessment

1000, pp. 155-176, Springer 2005

- 69 -

Kukkonen, J., Valkonen, E., Walden, J., Koskentalo, T., Aarnio, P., Karppinen, A., Berkowicz,

R. and Kartastenpaa, R. (2001). A measurement campaign in a street canyon in Helsinki and

comparison of results with predictions of the OSPM model. Atmospheric Environment 35, pp. 231-

243

Le Bihan, O., Gámez, A., Lohmeyer, A. and Berkowicz, R. (2004). Particulate emission and

dispersion in street canyon. In: R. Bercowicz, R. Britter and S. Di Sabatino, (eds.). Optimisation

of Modelling Methods for Traffic Pollution in Streets, TRAPOS, NERI, Denmark, pp. 60-66

Michail, A. (2003). Dispersion of pollution in urban areas. M.Sc.(Eng) Dissertation, Institute for

Transport Studies, University of Leeds

Moussiopoulos, N., Berge, E., Bøhler, T., de Leeuw, F., Grønskei, K.E., Mylona, S. and

Tombrou, M. (1996). Ambient air quality, pollutant dispersion and transport models. European

Environment Agency, Topic report No 19, January 1996

NAEI (2006). UK’s National Atmospheric Emissions Inventory website [online]. Available at:

http://www.naei.org.uk/ [Accessed 05/07/2006]

Namdeo, A. K. (1995). Modelling the Emission and Dispersion of Air Pollution from Motor

Vehicles. Ph.D. Thesis, University of Nottingham

NERI (2006). Description of the OSPM model. In National Environmental Research Institute

website [online]. Department of Atmospheric Environment, Denmark. Available at:

http://www.dmu.dk/International [Accessed 25th February 2006]

Nicholson, S.E. (1975). A pollution model for street-level air. Atmospheric Environment 9, pp.

19–31

Ntziachristos, L. and Samaras, Z. (2000). COPERT III - Computer programme to calculate

emissions from road transport, Methodology and emission factors (Version 2.1). European

Environment Agency, Technical report No 49, November 2000

Office for National Statistics (ONS) (2006). Environmental Accounts: Emissions;

Atmospheric, summary. ONS Website [online]. Available at:

http://www.statistics.gov.uk/default.asp [Accessed on 13th July 2006]

- 70 -

Ropkins, K. (2006). Vehicle Emissions Modelling. Summary paper, TRAN5700-Modelling

Traffic Pollution module, Institute for Transport Studies. University of Leeds, April 2006

Ropkins, K., Bell, M.C. and Tate, J. (2006). Introduction to the SRIF Instrumented Junction

Project. In: 38th Annual UTSG Conference, Dublin

Routesafe (2006). Routesafe UK Ltd website [online]. Available at:

http://www.routesafe.co.uk/ [Accessed on 20th July 2006]

Sharma, P. and Khare, M. (2001). Modelling of vehicular exhausts - a review. Transportation

Research - Part D 6, pp. 179–198

Tate, J. (2006). Emission Modelling: In Relation to Air Quality Studies. Draft paper, TRAN5700-

Modelling Traffic Pollution module, Institute for Transport Studies. University of Leeds, 6th

April 2006

Tsai, M.Y. and Chen, K.S. (2004). Measurements and three-dimensional modelling of air pollutant

dispersion in un Urban Street Canyon. Atmospheric Environment 38, pp. 5911–5924

Vardoulakis, S., Fisher B. E. A., Pericleous, K. and Gonzalez-Flesca, N. (2003). Modelling air

quality in street canyons: a review. Atmospheric Environment 37, pp. 155–182

Vardoulakis, S., Gonzalez-Flesca, N., Fisher B. E. A. and Pericleous, K. (2005). Spatial

variability of air pollution in the vicinity of a permanent monitoring station in central Paris.

Atmospheric Environment 39, pp. 2725–2736

Zannetti P. (1993). Numerical simulation modelling of air pollution: an overview. In: Zannetti, P. et

al. (eds.). Air Pollution. Computational Mechanics Publications, Southampton, pp. 3-14

- 71 -

Appendices

APPENDIX I: MANUAL TRAFFIC SURVEY FORM

DISSERTATION SURVEY Name: Date: Location: Northbound / Southbound

TIME CLASSIFIED COUNT

Start End Cars LGVs Rigid HGVs

Artic HGVs Buses

08:00 08:15

08:15 08:30

08:30 08:45

08:45 09:00

09:00 09:15

09:15 09:30

09:30 09:45

09:45 10:00

16:00 16:15

16:15 16:30

16:30 16:45

16:45 17:00

17:00 17:15

17:15 17:30

17:30 17:45

17:45 18:00

TOTALS

- 72 -

APPENDIX II: STATISTICAL ANALYSIS RESULTS

Wind direction dependence of measured CO concentrations for each

Streetbox. The vertical line indicates the direction for which the monitoring

point is in the windward side.

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 60 120 180 240 300 360

Wind Direction

Mea

sure

d C

O (m

g/m

3)

SB 125

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

0 60 120 180 240 300 360

Wind Direction

Mea

sure

d C

O (m

g/m

3)

SB 134

0.00

0.50

1.00

1.50

2.00

2.50

0 60 120 180 240 300 360

Wind Direction

Mea

sure

d C

O (m

g/m

3)

SB 146

0.00

0.50

1.00

1.50

2.00

0 60 120 180 240 300 360

Wind Direction

Mea

sure

d C

O (m

g/m

3)

SB 133

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Wind Speed (m/s)

Mea

sure

d (m

g/m

3)

SB 125

0.00

0.50

1.00

1.50

2.00

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Wind Speed (m/s)

Mea

sure

d (m

g/m

3)

SB 133

(degrees)

(degrees)

(degrees) (degrees)

- 73 -

Wind speed dependence of measured CO concentrations for each Streetbox.

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Wind Speed (m/s)

Mea

sure

d (m

g/m

3)

SB 134

0.00

0.50

1.00

1.50

2.00

2.50

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Wind Speed (m/s)

Mea

sure

d (m

g/m

3)

SB 146

Diurnal observed and predicted CO levels at East & West Roadside

East Side

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

00:00

01:00

02:00

03:00

04:00

05:00

06:00

07:00

08:00

09:00

10:00

11:00

12:00

13:00

14:00

15:00

16:00

17:00

18:00

19:00

20:00

21:00

22:00

23:00

Time of day

CO

(mg/

m3 )

SB134

SB133

R1

West Side 0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

00:00

01:00

02:00

03:00

04:00

05:00

06:00

07:00

08:00

09:00

10:00

11:00

12:00

13:00

14:00

15:00

16:00

17:00

18:00

19:00

20:00

21:00

22:00

23:00

Time of Day

CO

(mg/

m3 )

SB125

SB146

R2

- 74 -

0.40

0.60

0.80

1.00

1.20

1.40

1.60

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Wind Speed (m/s)

Mod

elle

d (m

g/m

3)

R2

Wind speed dependence of modelled CO concentrations for each receptor in WinOSPM.

Wind direction dependence of modelled CO concentrations for each receptor in WinOSPM

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

0 60 120 180 240 300 360

Wind Direction (degrees)

Mod

elle

d C

O (m

g/m

3)

R1

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Mod

elle

d (m

g/m

3)R1

s

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

0 60 120 180 240 300 360

Wind Direction (degrees)

Mod

elle

d C

O (m

g/m

3)

R2

- 75 -

APPENDIX III: fluidyn-PANACHE SETUP

Inserted background map of the urban area and user-defined road sources

User-defined pollutant sources (roads and urban area source) and monitoring points

- 76 -

3-D View of the examined domain in the urban area of Headingley.

The user-defined building outlines and heights are the obstacles inside the computational wind flow domain (Y-axis coincides with North).

modelling of air quality in street canyons · 2017-10-27 · modelling of air quality in street...

Documents