NILU TR 11/97

REFERENCE: Q-303
DATE SEPTEMBER 1997
ISBN 82-425-0913-1

Air Quality Monitoring

Systems and Application
Bjarne Sivertsen

NILU
Norsk institutt for luftforskning
Norwegian Institute for Air Research
Postboks 100 - N-2027 Kjeller - Norway
 3

Contents
Page

1. Introduction 7
2. A modern environmental monitoring and information system 8
2.1 The technical features of the system 8
2.2 Sensors and monitors 10
2.2.1 New instruments needed 10
2.2.2 Meteorological data 10
2.3 Environmental indicators 11
2.4 Data transfer and quality assurance 13
2.5 The data bases 14
2.5.1 The on-line data base 15
2.5.2 The emission data base 15
2.5 .3 Historical and background data base 16
2.5.4 Supporting data base 16
2.6 The models 17
2. 7 Data presentation; graphics and GIS 20
2.8 Environmental information to the public 21
3. Air Quality Indicators 23
3.1 The Conceptual Framework 23
3.1.1 Definition of Terms 24
3.2 Selected Air Quality Indicators (AQI) 25
4. Monitoring Programme 29
4.1 Programme design 29
4.2 Siting 30
4.2.1 Representativity 30
4.2.2 Sampling Station Density 31
4.3 Air quality measurement instrumentation 33
4.3.1 Passive samplers 33
4.3.2 Filter pack sampling 35
4.3.3 Glass filter sampling 35
4.3.4 Canister sampling 36
4.3.5 Adsorbent tubes 36
4.3.6 High volume PUP-sampler 36
4.3.7 Precipitation dust fall collection 37
4.3.8 Semi-automatic sequential samplers 37
4.3.9 Particulate matter sampling 40
4.3.10 Continuous automatic monitors .43
4.3.11 Open path measurements based on optical absorption 47
4.3.12 Meteorological measurements 48
4.4 Chemical analysis 49
4.4.1 SO2 analysis by the use of ion chromatography 49
4.4.2 SO2 analysis by the barium perchlorate-Thorin method 50

NILU TR 11/97
4

4.4.3 NO2 analysis 50

4.4.4 PM10 50
4.4.5 Lead 50
4.4.6 VOC analysis 50
4.4. 7 Analysis of (persistent) organic pollutants 51
4.4.8 Analysis of precipitation samples 51
4.5 Data retrieval and data handling systems 52
4.5.1 Data storage and transfer 52
4.5.2 Software 52
4.6 Quality Assurance (QA) 53
4.6.1 QA at site : 54
4.6.2 Network calibration 55
4.6.3 Routine controls at the reference laboratory 55
5. Meteorology 55
5 .1 The atmosphere 56
5.2 Large scale wind patterns 58
5.3 Terrain induced air flow 59
5.3.1 Wind and flow 59
5.3.2 Mountain and valley winds 60
5.3.3 Drainage winds 61
5.3.4 The three-dimensional circulation in mountainous regions 62
5.3.5 Sea and land breezes 64
5.3.5.1 Sea breeze 64
5.3.5.2 Land breeze 65
5.3.6 Deformation and separation of flow 65
5.4 Turbulence 69
5 .4.1 Mechanical induced turbulence 71
5.4.2 Thermally induced turbulence 72
5.5 Atmospheric stability 72
6. Air Pollution Modelling 75
6.1 Emission estimates 75
6.1.1 Emission from area sources 76
6.1.2 Emission from stationary point sources 77
6.1.2.1 Emission factors - point sources 77
6.1.3 Emissions from road traffic 80
6.1.3.1 Methodology 80
6.1.3 .2 Vehicle classes 80
6.1.4 The conception "emission factor" for road traffic 81
6.2 The emission inventory data base 82
6.3 Dispersion models 83
6.3.1 The Gaussian plume model 85
6.3.2 Traffic and car exhaust models 90
6.3.3 Puff trajectory models 94
6.3.4 Numerical models 94
6.3.5 Box models 95
6.3.6 Statistical models 95
6.4 Model applications 95
7. Data Presentation 101

NILU TR 11/97
 5

7 .1 Air pollution statistics 101

7 .2 Emission data 104
7.3 Meteorological data 106
7 .3.1 Measurements of wind speed and wind direction 107
7.3.2 Measurements of temperature 109
7 .3 .3 Atmospheric stability and turbulence 110
7.3.4 The combined wind-/stability matrix 112
7.3.5 Precipitation 113
7.3.6 The representativity of the wind measurements 114
7 .4 Air quality data 116
7.4.1 Trends, changes in time 116
7.4.2 Peak statistics 119
7.4.3 Spatial concentration distribution 120
7.4.4 Presentation of estimated concentration distributions 122
7.5 User friendly presentation 123
8. Impact assessment 125
8.1 The content of the environmental impact assessment (EIA) 125
8.2 Air Pollution Impact 126
8 .2.1 Air pollution and human health 126
8.2.2 Exposure estimates 127
8.2.3 Air pollution and flora and fauna 129
8.2.4 Air quality and atmospheric corrosion 131
8.3 Consequence analysis 131
8.4 Optimal abatement strategy planning 132
8.5 Cost/benefit analysis (example Manila) 134
8.5.1 Action plan 134
8.5.2 Future air quality for some abatement scenarios 135
9. References 136

NILU TR 11/97
6

NILU TR 11/97
 7

Air Quality Monitoring Systems and Application

1. Introduction
Development of technical monitors and telemetric systems have made environ-
mental data more readily available to planners, authorities and to the public. In
line with awareness and the strong focus on our environment the modem environ-
mental monitoring and surveillance systems have also become information
systems that can provide relevant information at different levels about the state of
the environment, quickly and precisely.

The integrated approach towards environmental management is based on the view

that the environment should be monitored and followed as an entity. This is also
in line with the concept "sustainable development" introduced by the Brundtland
Commission, and which has been widely adopted by both national governments
and international organizations.

Today's environmental information systems combine the latest sensor and

monitor technologies with data transfer, data base developments, quality assur-
ance, statistical and numerical models and advanced computer platforms for
processing, distribution and presenting data and model results. Geographical
Information Systems (GIS) are an important tool, particular for the presentation of
data.

These technologies can be used in environmental management to support

integrated pollution prevention and control. They can also be part of an emergency
management system to support actions and crisis management during emergencies
and accidents of various kinds. The content and operability of the system might be
quite different in the two cases. In the following we will describe the content of a
surveillance system for local and regional environmental management, for urban
areas or regions dealing with industrial problems, traffic, energy sources and solid
and liquid waste.

Most of the examples below are related to the development of a system for air
pollution monitoring and information. The examples given mostly apply to air
quality studies in urban areas. However, the descriptions can also very well be
applied to other types of environmental issues. Biological monitors or direct
impact monitoring ( on man and the environment) is not covered by the described
system.

NILU TR 11/97
8

2. A modern environmental monitoring and information system

2.1 The technical features of the system
The key features of the modem environmental information system is the
integrated approach that enables the user in a user friendly way to not only access
data quickly, but also use the data directly in the assessment and in the planning of
actions. The demand of the integrated system to enable monitoring, forecasting
and warning of pollution situations has been and will be increasing in the future.
The data may also be used for generating new indicators that relate directly to
health impacts. This will require that numerical dispersion models for air
pollutants are available with on-line data input as a part of the system in urban
areas.

Several systems are currently being developed and have been demonstrated in
selected areas in Europe. One such system, "ENSIS '94", an ENvironmental
Surveillance and Information System, was developed as part of the Eureka project
for the Winter Olympic Games in Lillehammer. (Sivertsen and Haagenrud, 1994).
The following description is based on this prototype.

The ENSIS concept has later been developed further into an AirQUIS module for
air pollution surveillance, a WaterQUIS module for water pollution, and similar
modules for noise, deterioration of materials and buildings etc. The different
modules are all operated under the same main framework and can be combined in
a flexible total system.

Other integrated systems are being established in Europe. One of the important
topics of the European Commission DG XIII Telecommunications, Information
Market and Exploitation of Research, Telematics Application Programme ( 1994-
1998) deals with this subject. Several major urban areas in Europe will thus be
involved in the establishment and demonstration of such systems.

The main features of the integrated surveillance and information system for the
environment is shown in Figure 1.

NILU TR 11/97
 9

A modern environmental
surveillance system
Background
Presentation
data
• Graphics
•GIS
Nois~ transfer
Data-
,----~Data-
~Water collecting base

·•tl~n
Buildings
,__-+ ~~~~It
LJ
Models

~
,I

Figure 1: The principal structure of a modern environmental monitoring and

information system.

An important objective for the modem environmental surveillance platform is to

enable on-line data and information transfer with direct quality control of the
collected data. This may require new sensor technology or modification of present
monitoring methods. Several monitors and sensors that makes on-line data
transfer and control possible are already available on the market. For several other
compounds and indicators this is not the case.

The system should include:

• Data collectors; sensors and monitors,

• data transfer systems and data quality assurance/control procedures,
• data bases included emission and discharge modules,
• statistical and numerical models (included air pollution dispersion models and
meteorological forecast procedures),
• user friendly graphical presentation systems including Geographical
Information Systems (GIS),
• a decision support system,
• data distribution systems and communication networks for dissemination of
results to "outside" users.

The different parts will in the following be described in more detail.

NILU TR 11/97
10

2.2 Sensors and monitors

2.2.1 New instruments needed
Modifications and development of new sensors and monitors are necessary to
establish a complete environmental information system that meets the
requirements of today's users. Several sensors and monitors for meteorology,
noise, air- and water quality are already available on the market. However, not all
of these can be linked on-line to a data transmission and data quality control
system.

A description of measurement techniques for environmental parameters are

presented in later chapters .. For air pollutants it is important to decide whether one
wants to measure in situ to obtain a point measurement or take an integrated
sample over a distance or a volume. In the latter case different optical methods
using light absorption have been developed and used during the last few years.
Specific methods including single line spectroscopy with advanced optical filters
or tuneable diode lasers emitting light at one particular wavelength have also
been, or are being developed for selected individual air pollutants. However, it is
difficult to obtain in situ measurements i.e. in streets with these instruments. The
cost of these instruments is also high, depending on the number of parameters
needed to get a good indication of the status of the air quality. (See indicators.)

A new generation of water quality sensors for process control and water
management was demonstrated during the ENSIS programme in Lillehammer
1994. It included the monitoring of drinking water, waste water treatment and
river water acidity.

2.2.2 Meteorological data

Meteorological data are important input data to a system that is to be used for
information, forecasting and planning purposes. Meteorological data are also
important for explanatory reasons together with climatological data.
Meteorological data are needed from the ground, normally collected along 10 m
towers, and up to the top of the atmospheric boundary layer. Automatic weather
stations are currently being used in most large field studies, in remote areas and in
complex terrain. Meteorological "surface data" such as winds, temperatures,
stability, radiation, turbulence and precipitation are being transferred to a central
computer via radio communication, telephone or satellite.

One of the more difficult parameter to obtain on a routine basis is the height of the
boundary layer as a function of time. This height is often related to and referred to
as the mixing height. When air quality models are being applied for exposure
modelling, information and forecasting and decision making purposes,
meteorological input data from the boundary layer are crucial.

NILU TR 11/97
 11

NILU automatic weather station

36m

\dT,6_10 C>e=--
1
L,(DD -dd)
N--
2

; ø ø

Every
5 min
online

Figure 2: A typical field monitoring station with an Automatic Weather Station

(A WS) and meteorological sensors along a 36 m tower.

To improve the meteorological input data for numerical air quality models in
urban areas, more advanced three dimensional wind and turbulence measurement
equipment should be included. These instruments can measure the atmospheric
turbulence directly. These turbulence data can be used directly to .estimate the
dispersion more accurately. Many areas have already installed Doppler sodar
systems that can measure the vertical structure of wind and turbulence. These data
are also subject to certain ambiguities, but represent a valuable additional input to
the models for on-line information and warning.

A combination of measurement data (at several locations) and model estimated

wind fields will represent the necessary input to numerical air pollution dispersion
models in a complex urban area. These models are usually set to estimate
concentration distributions on an hourly basis, and the most important parameters
are therefore the flow pattern and a correct picture of the transport of pollutants. In
some cases, especially when applying mesoscale and regional scale models,
remote sensing of weather systems from satellites may prove a useful tool for
estimating input data.

2.3 Environmental indicators

The selection of parameters included in the monitoring and model estimate
programme should enable an automatic access to data relevant for assessing the
environment included air pollution and atmospheric conditions, pollution of rivers

NILU TR 11/97
12

and seas, ground water, waste, noise and radiation. For all these environmental
compartments there should be a set of environmental indicators.

These indicators should represent a set of parameters selected to reflect the status
of the environment. An indicator may be a single variable of sufficient sensitivity
to reflect changes in the status of the environment. In some cases, however,
indicators may be derived from a set of independent variables in the system. The
selection of indicators should also allow evaluation of trends and developments.
The aim is that the indicators can form a basis for evaluating the impact on
humans and the environment as a whole and thereby be relevant for information,
warning and decision making purposes.

Many national and international authorities are presently working with processes
to select environmental indicators. The result of this work will not be available in
another few years. In the meantime, for air quality, the selected parameters are
mostly related to air pollutants for which air quality guideline values are available.

The development of environmental indicators in Europe will contribute to the

harm onization of several initiatives. This activity will be important input to the
design and content of monitoring programm es. Harmonization is an important
concept both in monitoring and in modelling. It allows different methods to be
used to measure the same variable to predetermined levels of accuracy and
precision. Even if different methods are applied the data from each location can be
comparable and compatible.

The selected set of environmental indicators will be used by local and regional
authorities as a basis for the design of measurement programmes and for reporting
the state of the environment.

The establishment of environmental indicators will help to:

• identify the quality of the environment,
• quantify the impact,
+ harmonize data collection,
+ assess the status and the rate of improvement/deterioration,
+ identify needs for and support the design of control strategies,
+ support input to management and policy changes.

The indicator should represent the "pressure" on the environment and include both
background indicators and stress indicators. So-called response indicators are
selected to reflect the societies awareness or response to its surroundings.

The indicator should:

+ be relevant in connection with environmental quality,

• be easy to interpret,
• respond to changes,
+ provide international comparisons,
+ have a target or threshold value that provides a basis for assessment,
+ be able to show trends over time.

NILU TR 11/97
 13

It should also be possible to measure with reasonable accuracy. It should be

adequately documented and linked to public awareness; health impact, building
deterioration, vegetation damage etc. Selected indicators should respond to
mitigation actions to prevent human made negative impacts on the environment.

Indicators might also be aggregated data and not necessarily observed single
parameters. The modern environmental surveillance and information systems
(ENSIS) include good quality on-line meteorological data, numerical dispersion
models with emission inventories. These models are capable of estimating
concentration distributions on an hourly basis. These distributions can be linked to
population distribution maps, building material inventories, vegetation maps etc.
to give exposure estimates.

These aggregated, estimated data will express directly the impact and stress to the
environment (health, materials, vegetation) and will in the future represent a better
indicator for international comparisons and trend analyses. It will also represent an
improved measure for the actual air pollution problem in a given (well defined)
area or region.

2.4 Data transfer and quality assurance

Specially designed data loggers for environmental data are available. Data loggers
designed and built by NILU were included in the ENSIS '94 application. The
logger should be robust and serve as a local backup storage unit in case of link
brake down (lightening, storms etc.). The logger is directly linked to a modem.

Data transfer can be via local radio communication for limited distances. This has
been the case for a distributed local net of several meteorological stations where
data are transmitted via radio link to the main station in the area. Data will further
be transmitted on public telephone lines or via satellite to the main computer
facility. The central unit might be a major field station or a central laboratory. For
an emergency system developed for the Eureka project MEMbrain, a field
laboratory has been established with a work station computer including all
modelling tools. (Sivertsen, 1994b)

Data quality assurance programmes including direct quality control is performed

at different levels in the data collection process;

• in field during automatic and manual calibrations and controls,

• at the central data collection base following quality assurance routines as de-
scribed i.e. in ISO 45001 from the International Standardization Organization,
• in approvals to the final data base,
• through simple statistical and graphical evaluations to check validity and
representativeness of data.

NILU TR 11/97
 14

The quality control procedures give the data credibility. The data become reliable,
which is essential when using the data for reporting, controls and planning. To be
used with confidence for scientific and environmental management purposes the
data must also be comparable and compatible.

Integrated data from local sites and from various environmental compartments
require comparable data quality. The various local networks have to operate to
high standard including proper implementation of good practice by network
managers and site responsible personnel.

2.5 The data bases

The development of an associated data base or metadata is important to all
modern environmental monitoring and information systems. The data base system

~ Information
._.______,'--" and planning

Measure- Other data

Emission demographic
ments buildings
land use

Models

~r-frien~r"lii
Contour plots Time series Tables

Figure 3: The associated data bases are linked to a modelling system which pro-
vides user friendly presentations of all kinds of information from the
system.

NILU TR 11/97
 15

may consist of several data bases which serve as main storage platforms for:

• on-line collected environmental data,

• emission and discharge data included emission modelling procedures,
• historical data and background information like area use, population distribu-
tions and trends,
• regulations, guideline values and information on the support and decision
making process.

The data bases contain information that enables an evaluation of the actual state of
the environment and it includes data for establishing trend analyses, warnings and
the undertaking of countermeasures in case of episodic high pollution.

2.5.1 The on-line data base

All data collected on-line will after quality assurance and controls be part of the
information data base. From this base it will be possible to obtain quick graphical
presentations, or to subtract data for public information purposes etc.

2.5.2 The emission data base

The emission data base is an interactive platform for collecting input data for
emission estimates. It contains information about the sources, emission factors,
consumption data, information on locations (gridded co-ordinates), stack heights,
stack parameters, fuels etc. The emission data base can be operated directly by the
user, who can use the emission models directly to present emission data directly.
Any changes and additions to the emission data base will result in updated
emission estimates with links to the dispersion models and resulting database for
graphical presentation.

NILU TR 11/97
 16

N02
1---+---+--+--+-++--+--+---+--+---1f--+---++--+--+---1f--+--+----H Emissions
1---+---+--+--+-1--+--+--+----++--+---1f--+---++--+--+---1f--+--+----H Oslo

Figure 4: An emission inventory of NOx emissions presented in a 1 xi km grid

for Oslo. The emission estimates have been based upon fuel consump-
tion data, industrial sources, traffic and emission factors.

2.5.3 Historical and background data base

The historical and background data base module includes relevant objects and
information such as monitoring stations and sensors, sensor developers, respon-
sible institutions, locations and measurement schedules, methods, data owners,
maintenance routines etc. It also contains information about earlier and additional
environmental data collected in the area. Background information such as area
use, population distributions and inventories of vegetation and materials/buildings
in the area may be an important part of this data base. Such information can be
used for impact assessment estimates and for some of the emission estimates.

2.5.4 Supporting data base

The supporting data base, which may be part of the background data base contains
information on regulations, requirements, air quality guideline values or water
quality standards for various applications.

Information about regulations and plans given by local authorities or by

governmental bodies should be included in this database, as well as support
actions and emergency procedures.

NILU TR 11/97
 17

The total associated database system will also serve as a link to a meta
information system which includes information on external environmental data.
These functions might also include:

+ navigation facilities to access the needed information,

+ support for standardization activities,
+ world wide web/internet functions and bridges.

The data base model is designed to support local and regional levels and meets
most of the requirements specified by the users.

Modifications and additions must be easily made in the database. Routines for
safety copying and reconstruction must be available. Different data deliveries
might be operating in different systems. This requires the establishment of
different communication systems with open communication solutions.

2.6 The models

In the modem multi compartment environmental information system (like ENSIS)
steps have been taken to establish models for air pollution dispersion, for water
quality and noise and for other environmental impact assessment estimates.
Models for these media will be essential when the programmes are to be used for
planning purposes.

The air pollution dispersion models are a well-established and fully implemented
part of the system. These models have been tested and demonstrated as part of the
integrated surveillance systems and is presently being operated in several cities on
a routine basis. Also water quality modelling is available and is being tested and
verified as part of the EN SIS system.

Different types of dispersion models have been developed and applied to estimate
the ambient impact of air pollution emissions from point-, line- and area sources.
These will be described in more details in ch. 6.

The selection of models to be used in a specific case is dependent upon the spatial
and temporal scales, complexity of source configurations and chemistry,
topographical features, climate and instationarity/inhomogeneity in the
meteorological conditions of the area. It is advisable to consult experts in this
process.

A variety of different models are available on the market today. However, one
should note that it may be a significant step from obtaining a model to actually
having an operable modelling tool for a specific area and application.

Different types of models available are taken from the air pollution surveillance
programmes. They range from single quasi stationary Gaussian type single source
models based upon analytical solutions of the mass balance equations, to
advanced numerical models which require large computers.

NJLU TR 11/97
18

The simplest models can be used on personal computers for impact assessment.
These models can estimate 1 h average concentration distributions downwind
from ground level, diffusive and elevated single sources. (Sivertsen 1980, Bøhler
1987)

One step up represents the short term model for estimating 1 h average
concentration distributions for emissions from multiple source industrial
complexes (Bøhler 1987). This includes the multiple source Gaussian type models
for estimating short term or long term integrated concentrations in a gridded co-
ordinate system. Two different type of such models have been developed at NIL U;
CONDEP for monthly, seasonal and annual average concentration distribution
estimates (Bøhler 1987) and KIL DER which is a flexible emission inventory
linked to multiple source Gaussian type dispersion models for line, area and point
sources. (Gram and Bøhler 1992).

Episode
model
Lillehammer
22 Feb 1994
2200 h
N02 (µg/m3)

Figure 5: Modelling of one hour average N02 concentration distributions from

the Lillehammer Winter Olympic programme.

The grid system used by the models is specified by the user to match the specific
problem and the area considered. The resolution, grid spacing and total area can
easily be modified and changed depending upon the specific needs.

These models need as input data some background information on;

+ source characteristics and emission data,

+ area characteristics (surface roughness, topography etc.),
+ measurement data (measurement type, heights etc.),
+ meteorological data (wind, stability, mixing height, temperatures etc.),
+ dispersion coefficients (type to be used and parameters),
+ dry and wet removal coefficients,
+ location of receptor points (distances or grid specifications).

NILU TR 11/97
 19

All the NILU models have been well documented and are being used for planning
purposes and for impact assessments both nationally and internationally.

Small scale models are also available for estimating the air pollution load from
traffic in street canyons and along roads. A commercially available model,
ROADAIR (Larssen and Torp, 1993), estimates emissions, concentrations and
exposure along the road system based upon traffic data. These input data may
originate from traffic models or from traffic density data and on-line traffic
counting.

On a spatial scale from about 1 to 100 km there are several types of numerical
models available; both Lagrangian type and Eulerian type models. The
Lagrangian type models follow puffs of air pollutants estimating in each puff the
turbulent diffusion, chemical reactions and deposition processes. The turbulence
description and the diffusion processes may be treated in different ways.

One example is the INPUFF model (Knudsen and Hellevik, 1992) which is based
upon Gaussian concentration distributions in the puff. This model also includes
chemical and physical reactions and processes. Another model of this type is the
Danish operational puff diffusion model RIM PUFF (Mikkelsen et al., 1987). This
model was developed by Risø National Laboratory to provide risk and safety
assessment in connection with e.g. nuclear installations.

One example of an Eulerian type numerical dispersion model is the EPISODE

model developed by Grønskei et al. (1993). This is a time-dependent finite
difference model normally operating in three vertical levels, combined with a puff
trajectory model to account for subgrid effects close to individual sources. When
the size of the puffs reaches the horizontal and vertical grid size the transport and
dispersion is treated as a numerical box model. The mass of pollutants are then
added to the average value for that grid element. The model can thus treat point
sources, area/volume sources and line sources. The wind field used as input to the
model may be homogeneous or inhomogeneous for each time step dependent
upon the meteorological input data available.

For the selection of models to be used in a specific case there have been different
methods indicated. Sivertsen (1979) indicated a flow chart for selecting models
dependent upon type and complexity of the sources, spatial and temporal scales,
chemical composition (secondary or primary pollutants), topographical features,
climate and meteorological features of the selected area.

For further information on the use of models, Hanna et al. (1982) give a good
overview of the topic. One important issue when using dispersion models is to
obtain adequate meteorological input data. Meteorological pre-processors have
been developed during the last few years to handle this problem. (Paumier et al.,
1985 and Bøhler et al., 1996). These pre-processors can estimate meteorological
dispersion and the basic meteorological variables of interest for diffusion
modelling based upon the current concepts regarding the structure of an idealized
boundary layer. (Gryning et al., 1987). Methods are also provided for estimating

NILU TR 11/97
20

the vertical profiles of wind velocity, temperature and the variances of the vertical
and lateral wind velocity fluctuations.

2.7 Data presentation; graphics and GIS

Environmental data collected through the automatic monitoring and telematic
network will be quality controlled and transferred for storage in the integrated
relational databases. Statistical programmes for control of quality and
representativeness will be used, and the first results can within one hour after field
collection be presented using user-friendly graphical tools.

The information may be multimedia: texts, tables, graphs, images, sound or video
dependent on the end user. The presentations have to be designed to meet the user
needs. These users may be:

+ authorities at different levels (municipal, regional, national, international),

+ industrial users,
+ schools, universities and the scientific community,
+ various organisations,
+ the public and media.

The environmental data are usually linked to geographical sites. In particular

when monitoring data are supported and supplied by model estimates of spatial
concentration distributions and impacts, it is suggested that the presentation of the
results would involve the use of maps or digitalized Geographical Information
Systems (GIS).

Geographical information systems based on advanced raster/vector technology has

been developed to handle maps, networks, symbols and various objects. They can
handle both geographical information and technical documentation and present
this in graphical form. The basic raw map information has normally been work-
station based, but user friendly PC based applications for displaying e.g.
environmental data have been developed during the last few years.

The GIS user can easily organise selected data from various data bases. Tematic
maps can be produced combined with time series graphical presentations and
results from model calculations. The system will display the results of planned
actions based upon simulation models and thus act as a more user friendly
decision support system.

For the application of ENSIS during the Winter Olympics in 1994 Arclnfo and
ArcView were selected as the map reference systems. The GIS tool was directly
linked to the data bases, from which statistical evaluations, graphical presentations
and spatial distributions from numerical models were presented.

NILU TR I 1/97
 21

2.8 Environmental information to the public

A wider distribution of environmental data to the public has become a part of the
development of modern environmental surveillance and information systems.
New approaches have been developed for dissemination of environmental
information which can be adapted to different information distribution systems.
These systems could be teletext, public telephone network, special designed health
advice information lines, telefax distributions, INTERNET networks etc ..

Information of air quality in urban areas have been issued to the public on a daily
basis described in terms of "very good", "good", "poor" etc. Some European cities
already provide this type of information. The modem information system will
focus more on variable messages and more updated access to the data through
teletext or Internet applications.

As part of the ENSIS development a windows-based PC presentation solution was

developed giving multiple access to different databases meeting common
graphical user interfaces. It is important that the platform is graphical and
preferably MS-Windows or X-Windows operating systems in a client-server
network configuration, that can provide access via wide area networks (WAN) to
external databases.

Open communication solution

The Data
User Presentations
> 35
operativ
systems

EDA/SQL 1<.:~=C>I

Spreadsheets
Tables Over 35
Data base
Reports platforms
Graphics
GIS Over12
communication
protocols

Figure 6: The user oriented open communication solution established during the
Eureka ENSIS development project. Any type of data could be
accessed and presented through a flexible graphical user interface
based on Windows 3.1.

NJLU TR 11/97
22

Several local authorities in Norway can presently obtain air quality information in
graphical form from several urban areas participating in the national surveillance
programme co-ordinated by the Norwegian Pollution Control Authorities. In Oslo
and Bergen this system is being used to develop information and forecasts on air
quality to the public. Lines have been set up to an information screen available for
the public and information is also being issued in the media daily.

NILU TR I 1/97
 23

3. Air Quality Indicators

It is normally not possible to measure all the air pollutants present in the urban
atmosphere. We therefore have to choose some indicators that should represent a
set of parameters selected to reflect the status of the environment. They should
enable the estimation of trends and development, and should represent the basis
for evaluating human and environmental impact. Further, they should be relevant
for decision making and they should be sensitive for environmental warning
systems.

Many national and international authorities are at present working with processes
to select environmental indicators. The selected parameters for air quality are
strongly related to air pollutants for which air quality guideline values are
available. The interrelationships between the indicators and other related
compounds, may, however, vary slightly from region to region due to differences
in emission source profiles.

The selected set of environmental indicators are being be used by local and
regional authorities as a basis for the design of monitoring and surveillance
programmes and for reporting the state of the environment.

Air quality indicators should

• provide a general picture,

• be easy to interpret,
• respond to changes,
• provide international comparisons,
• be able to show trends over time.

Measurement techniques should be reasonably accurate and within an acceptable

cost. The effect of indicators on health impact, building deterioration, vegetation
damage, etc., should be adequately documented and linked to public awareness.
Selected indicators should respond to mitigation actions to prevent manmade
negative impacts on the environment.

3.1 The Conceptual Framework

In general terms, an indicator can be defined as a parameter, or a value derived
from parameters, which provides information about a phenomenon. The indicator
has significance that extends beyond the properties directly associated with the
parameter value. Indicators possess a synthetic meaning and are developed for a
specific purpose. This points to two major functions of indicators:

1. They reduce the number of measurements and parameters which normally

would be required to give an "exact" presentation of a situation;
2. They simplify the communication process by which the results of measurement
are provided to the user.

NILU TR 11/97
24

3.1.1 Definition of Terms

INDICATOR
A parameter, or a value derived from parameters, which points to,
provides information about, describes the state of a phenomenon/en-
vironment/area, with a significance extending beyond that directly
associated with a parameter value.

INDEX
A set of aggregated or weighted parameters or indicators.

PARAMETER
A property that is measured or observed.

INDICATORS OF ENVIRONMENTAL PRESSURES

Correspond to "pressure" box of PSR framework. They describe
pressures on the environment caused by human activities.

INDICATORS OF ENVIRONMENTAL CONDITIONS

Correspond to "state" box of the Pressure State Response framework.
They comprise environmental quality and aspects of quantity and quality
of natural resources.

RESPONSE INDICATORS
Correspond to "response" box in PSR framework. In the present context,
the word "response" is used only for societal (not ecosystem) response.

INDICATORS FOR USE IN PERFORMANCE EVALUATION

Selected and/or aggregated indicators of environmental conditions,
indicators of environmental pressures and indicators of societal responses
for the purpose of environmental performance evaluation.

ENVIRONMENTAL INDICATORS
All indicators in the Pressure State Response framework, i.e. indicators
of environmental pressures, conditions and responses.

As indicators are used for varying purposes it is necessary to define general

criteria for the selection of indicators. Three basic criteria have been used in
OECD work: policy relevance, analytical soundness and measurability (OECD,
1994).

In large parts of its work, the Group on the State of the Environment uses the
Pressure State Response (PSR) framework. The PSR framework (Figure 7) 1s
based on a concept of causality:

+ Human activities exert pressures on the environment and change its state : i.e.
quality and the quantity of natural resources.
+ Society response to these changes through environmental, general economic
and sectoral policies.

NILU TR 11/97
 25

~---P_
r_e_
ss_u_r_
e ~I ~I s_t_
a,_e__
Information
~I ~I R_e_s_
po_n_s_e _,

Human Activities State of the Economic and

Environment Information Environmental
and the of Agents
Energy Pressures Natural Resources
Transport Administrations
Industry Air Households
Agricultre Resources Water Enterprises
Others Land Societal
Living Resources Responses International
(Decisons
Actions)

Societal Responses (Decisions - Actions)

Figure 7: Pressure State Response Framework.

3.2 Selected Air Quality Indicators (AQI)

Air quality indicators have been selected for different environmental issues and
challenges. Not all indicators are specific enough to address only one issue. The
nature of air pollution involve that some indicators address several issues. Some
of the issues that have to be addressed are

• climate change,
• ozone layer depletion,
• acidification,
• toxic contamination,
• urban air quality,
+ traffic air pollution.

As can be seen from the list the indicators have to cover all scales of the air
pollution problems (in space and time) to address different type of impacts and
effects.

In Europe different indicators have been established for characterizing different air
pollution types, as examplified in Table 1. (Sluyter, 1995)

NJLU TR 11/97
26

Table 1: Indicators selected for different types of air pollution in Europe. The
number of cities in Europe where given Air Quality Guideline (AQG)
values are exceeded are given. ( Sluyter, 1995)

Pollution type Indicator AQG (µg/m3) Cities with Effects

observed
exceedances
(%)

Short term effects

Summer smog 03 150-200 (hour) 84 Lung function de-
crements, respira-
tory symptoms
Winter smog SO2+PM 125+ 125 (day) 74 Decreased lung
function; increased
medicine use for
susceptible children
Urban traffic N02 150 (day) 26

Long term effects

Traffic/industry Lead 0.5-1.0 (year) 33 Effects on blood
formation, kidney
damage; neurologic
cognitive effects
Combustion SO2 50 (year) 13 Respiratory
symptoms,
PM 50 (year) 0 chronic respiratory
illness

The most commonly selected air quality indicators for urban air pollution are
carbon monoxide (CO), nitrogen dioxide (N02), sulphur dioxide (S02), particles
with aerodynamic diameter less than 10 µm (or 2,5 µm), PM10 (PM2.5) and ozone.

Some selected air quality guideline (AQG) values for these indicators are
presented in Table 2:

Table 2: Typical air quality guideline (AQG) values for some selected indi-
cators based on impact on public health (WHO, 1987 and 1995)

Indicator AQG (µg/m3)

averaoinc time
1 h 24 h Year

CO (mg/m3) 100 10 (8 h) -
NO2 (µg/m3) 200 40-50
SO2 (µg/m3) 500 125 50
PM10 (µg/m3) - 70** -
Black Smoke* 125 50 ... -
Ozone (µg/m3) 150-200 120 -
* Together with SO2
** Norway (SFT)
*** 8 h average (1995 recommend.)

NILU TR 11/97
 27

The most important indicators when discussing health impacts especially linked
to respiratory hypersensitivity are considered to be oxidized pollutants such as
NO2 and ozone. SO2 combined with acid aerosols are also associated with
respiratory problems. For particulate matter the particle size plays an important
role. Primarily the fine fraction ( <2,5 µm) of particles, often associated with
strong aerosol acidity or sulphates or correlated with gaseous components, is
assumed to impact the respiratory system.

It should also be noted that a common feature of exposure to the primary

compounds NO2, SO2, ozone and PM (particulate matter) id that the resulting
health effects may be altered in the presence of other compounds anci/or
aeroallergens. The interaction of the compounds can be synergistic. These
considerations are generally not taken into account when AQG values are
established.

Although the AQG take into account the most sensitive populations, known or
supposed interactions with climatic factors are not accounted for. The existence of
a threshold value has not necessarily been documented for all compounds. For
compound where this is the case there is normally a safety margin between the
lowest known effect and the AQG value.

Peak statistic bar charts have been produced for acute health effect indicators for
each criteria pollutant and the annual mean lead concentration. An example of this
is presented in chapter 7. The indicators for which bar charts have been elaborated
are shown in Table 3.

Table 3: Indicators for elaboration of air quality status in OECD cities

Pollutant Unit Indicator

Carbon monoxide mq/m3 Annual max. 8-hour runninq averaqe

Nitrogen dioxide µg/m3 Annual average

Ozone µg/m3 Annual max. t-hour average

Annual max. 8-hour runninq averaqe

Particulate matter µg/m3 Annual max. 24-hour average

Sulphur dioxide µg/m3 Annual max. t-hour average

Annual max. 24-hour averaqe

Air pollution concentrations in OECD cities have been compared to WHO

Guideline criteria which enables assessment of the likely impact of the air quality
upon health. Some of the indicators adopted by the OECD have been discontinued
in the revised WHO Guidelines.

Recently WHO has presented new proposed air quality guidelines for protection
of terrestrial vegetation. These proposals are presented in Table 4.

NILU TR 11/97
28

Table 4: New proposed Air Quality Guideline values presented to protect

terrestrial vegetation (WHO, 1995)

Substance Guideline Averaging Remarks

value time
µg/m3

Nitrogen dioxide 95 4 hours In the presence of S02 and 03

levels which are not higher than
30 µg/m3 (arithmetic annual
average) and 60 µg/m3,(average
during growing season
30 1 year
Total nitrogen Sensitive ecosystems are
deposition 3 q N/m2 1 year endanqered above this level
Sulphur dioxide 30 1 year Insufficient protection in the case
of extreme climatic and topo-
graphic conditions.
100 24 hours
Ozone 65 24 hours Vegetation
60 100 days Growinq season
Peroxyacetylnitrate 300 1 hour Vegetation
80 8 hours Veqetation

NILU TR 11/97
 29

4. Monitoring Programme
4.1 Programme design
As part of the establishment of an air quality monitoring and surveillance system,
a programme has to be established to design and plan the details and content of
such a system. This programme should be undertaken including the following
topics:

1. Define the objectives and strategies for the measurement programme,

2. define the contents,
3. perform a screening,
+ problems and relevant air pollution sources,
+ collect available data (meteorology and air quality),
4. evaluate existing data,
+ representativeness equipment,
+ QA procedures,
5. plan the programme in detail,
+ siting studies,
+ consider field investigations,
+ emission inventorying, simple modelling,
+ select relevant sites,
6. optimize measurements, (cost/effective design),
7. procure instruments,
+ specify technical requirements,
8. establish and initiate operation,
+ laboratory control systems,
+ develop standard operational procedures (SOP),
+ define and describe QA/QC procedures,
9. training.

A general objective for the air quality measurement programme (monitoring,

sampling and analysis) is often to adequately characterize air pollution for the area
of interest, with a minimum expenditure of time and money. The measurement
and sampling techniques to be used in each case will be dependent upon a
complete analysis of the problem. The main objectives stated for the development
of an air quality measurement and surveillance programme might be:

1. Background concentrations measurements,

2. air quality determination to check,
+ air quality standards to monitor current levels,
+ to detect individual sources,
+ to collect data for land use planning purposes,
3. observe trends (related to emissions),
4. develop abatement strategies,
5. assess effects of air pollution on health, vegetation or building materials,
6. develop warning systems for prevention of undesired air pollution episodes,
7. research investigations,
8. develop and test diffusion models,
9. develop and test analytical instruments.

NILU TR I 1/97
30

Once the objective of air sampling is well defined, a certain operational sequence
has to be followed. A best possible definition of the air pollution problem together
with and analysis of available personnel, budget and equipment represent the basis
for decision on the following questions:

1. What spatial density of sampling stations is required?

2. How many sampling stations are needed?
3. Where should the stations be located?
4. What kind of equipment should be used?
5. How many samples are needed, during what period?
6. What should be the sampling (averaging) time and frequency?
7. What other than air pollution data are needed:
+ meteorology,
+ topography,
+ population density,
+ emissions,
+ effects, etc.?
8. What is the best way to obtain the data (configuration of sensors and
stations)?
9. How shall the data be communicated, processed and used?

The answers to these questions will vary according to the particular need in each
case. Most of the questions will have to be addressed in the siting studies
discussed in the next chapter.

4.2 Siting
4.2.1 Representativity
It is important to bear in mind, when measuring air quality or analyzing results
from measurements, that the data you are looking at is a sum of impacts or
contributions originating from different sources on different scales.
The total concentration is a sum of

+ a natural background concentration,

+ a regional background,
+ a city average background concentration (kilometre scale impact),
+ local impact from traffic along streets and roads,
+ impact from large point sources; industrial emissions and power plants.

To obtain information about the importance of these different contributions it is

therefor necessary to locate monitoring stations so that they are representative for
the different impacts. This normally means that more than one monitoring site is

NILU TR 11/97
 31

needed for characterizing the air quality in the urban area. It is also important to
carefully characterize the monitoring representativeness, and to specify what kind
of stations we are reporting data from. An often used terminology is

+ urban traffic,
+ urban commercial,
+ urban residential and
+ rural sites.

When considering the location of individual samplers, it is essential that the data
collected are representative for the location and type of area without undue
influence from the immediate surroundings.

In the design of an urban air quality monitoring programme the following general
guidelines should be considered:

+ All stations (air intake) should be located at the same height above the surface,
a typical elevation in residential areas is 2 to 6 m above ground level.
+ Constraints to the ambient airflow should be avoided by placing the air intake
at least 1,5 meters from buildings or other obstructions.
+ The intake should be placed away from microscale or local time varying
sources.

4.2.2 Sampling Station Density

The number of stations needed to answer the objectives of the arr pollution
sampling, depends on many factors such as

+ types of data needed,

+ mean values and averaging times,
+ frequency distributions,
+ geographical distributions,
+ population density and distribution,
+ meteorology and climatology of the area,
+ topography and size of area,
+ location and distribution of industrial areas.

A rough indication of the minimum number of sampling stations needed as a

function of population density is given in Figure 8. for a typical community air
quality network. Automatic continuous sampling equipment in general involve
fewer stations than an integrating sampling device (24 hr average or more).

NTLU TR 11/97
32

Monitoring siting
Number of stations in urban areas
10000~----~-,.,--,,---~~-~---------.--------~~
,,,~ ,,-"
/Average
--
/ Average
5000 t,.-Region ,- Region
, /
l Minimally --
Highly Polluted Region
,' Polluted ,,-
l Region ,,'

2000

1000 I
I
I
I
I

I
I
I
//
I

~ I
I
I
l
C:
(ll I I
I I
~ I
I
I
I
0 500 I
~ I
I
.s I
I

c:· I
I
I
Meehan i cal-integrated
~ I
1 200 (continuously operating)
& (TSP, SO2)
100 Automatic- 1 km
-,
continuous
50
(SO2, CO, HC,
NOx, Oxidants etc)

20

Number of Stations
10-UJ'-'---------L----------------------~
5 10 15 20 25 30 35 40 45 50 55 60

Figure 8: Minimum number of measurement stations needed for a typical

community air quality measurement network.

The selection of sampling time is a function of the air pollutant characteristics

(emission rate, life time) and time specifications of the air quality criteria.

The ability of combining the air quality data with meteorological data through
dispersion modelling, also is a very important tool in the design of sampling
networks.

If the location of the maximum air pollution area is known from a limited
information about the region's meteorology, and the only objective is to check
that air quality standards are met, in some cases even one sampling station may be
sufficient.

NILU TR 11/97
 33

In a topographical complex area with hills, valleys, lakes, mountains etc., there
are considerable local spatial and temporal variations of the meteorological
parameters, and thus the dispersion conditions. To answer the same questions,
more sampling stations are needed in such areas than in flat homogeneous terrain.

Typical for a flat area is also that spaced stations (as proposed by the German
Federal regulations or as is the New York City's aerometric network) average out
spatial variations and thus can give net results representative for the area as a
whole.

4.3 Air quality measurement instrumentation

Instruments for measurements of air pollutants may vary strongly in complexity
and price from the simplest passive sampler to the most advanced and most often
expensive automatic remote sampling system based upon light absorption
spectroscopy of various kind. The following Table indicate four typical types of
instruments, their abilities and prices.

Table 5: Different types of instruments, their abilities and price.

Instrument Type of data Data availability Typical Typical price

type collected averaging (US$)
time
Passive Manual, in situ After lab analyses 1-30 days 10
sampler
Sequential Manual /serni- After lab analyses 24 h 1 000
sampler automatic , in situ
Monitors Automatic Directly, on-line 1h >10 000
Continuous, in situ
Remote Automatic/Continua Directly, on-line <1 min >100 000
monitoring us, path integrated
(space)

Relatively simple equipment is usually adequate to determine background levels

(for some indicators), to check Air Quality Guideline values or to observe trends.
Also for undertaking simple screening studies, passive samplers may be adequate.
However, for complete determination of regional air pollution distributions,
relative source impacts, hot spot identification and operation of warning systems
more complex and advanced monitoring systems are needed. Also when data are
needed for model verification and performance expensive monitoring systems are
usually needed.

4.3.1 Passive samplers

Simple passive samplers have been developed for surveillance of time integrated
gas concentrations. These type of samplers are usually inexpensive in use, simple
to handle and have an adequate overall precision and accuracy dependent upon the

NILU TR 11/97
34

air pollution concentration level in question. This method has been used m
industrial areas, in urban areas and for studies of indoor/outdoor exposures.

25
Passive vs. active NO2 sampling
Plastic tube 50
mm
45
....•··
40
....•··· 0

0 •••

35 ···'/
0 30 "o . .,...-;; V
z
Pre filter Gl 25 ~ ....;;;; V 0

>
:.::;
(..) 20 ._-9'.,,, ~
< 0
d'
,. 0

15
10 J -Q

5 /" 0

0 /
0 5 10 15 20 25 30 35 40 45 50
Passive NO 2concentrations

Figure 9: a) A passive impregnated filter sampler based on molecular diffusion.

b) The integrated passive sampling of S02 and NO2 is well correlated
with available active sampling methods

A sensitive diffusion sampler for sulphur dioxide (SO2) and nitrogen dioxide
(NO2) developed by the Swedish Environmental Research Institute (IVL) and has
been used in several investigations by NILU to undertake a screening of the
spatial concentration distribution in ambient air.

The sampler includes an impregnated filter inside a small plastic tube. To avoid
turbulent diffusion inside the sampler, the inlet is covered by a thin porous
membrane filter. Gases are transported and collected by molecular diffusion. The
uptake rate is only dependent upon the diffusion rate of the gas. The collection
rate is 31 l/24h for SO2 and 36 l/24h for NO2. Also NH3 can be collected at a rate
of 591124h.

For SO2 the measuring ranges are approximately 0, 1-80 ppb for a sampling period
of one month. The corresponding range for NO2 is 0,02-40 ppb. The passive
samplers are assembled and made ready for use at NILU .After exposure the
samplers are usually returned to NILU where concentrations of SO2 are
determined as sulphate by ion chromatography. NO2 and NH3 is determined by
spectrophotometry.

The passive samplers have been used in several field studies to map concentration
distributions, both as part of a screening to identify the magnitude of the problem
and for modelling purpose to estimate total emission rates and possible impacts.
The NO2 concentration distribution in Oslo on a winter day is only one example
shown in Figure 10.

NILU TR 11/97
 35

Oslo, 3-4 Fetv4 9

0j
2 conce
apping, Passiv

Figure JO: The 24 h average N02 concentration distributionfor Oslo measured

on 3 -4 February 1994 with 20 passive samplers show that the highest
concentrations occurred along the main road systems and in central
parts of down-town Oslo. This 24 h average distribution may change
considerably from one day to the next depending on meteorological
conditions in the Oslo airshed.

4.3.2 Filter pack sampling

The filter pack for air sampling consists of a filter holder with a teflon pre-filter
for particles and two impregnated paper filters for gases. The filter holder is
connected to a pump with flow controller which pull a steady airflow through the
filters. The detection limit is better than for the other methods but the method is
more labour intensive and is dependent of extra sampling equipment such as a
high precision electric pump.

4.3.3 Glass filter sampling

The Glass filter sampler consists of a glass bulb with a impregnated glass filter
inside. The glass bulb is connected to a calibrated pump that draws a steady
airflow through the filters. After exposure the glassbulb is sent to the laboratory
for analysis, then the filter is washed and used again. The detection limit is better
than for the other methods but the method is more labour intensive and depends of
extra sampling equipment such as a high precision electric pump.

NILU TR 11/97
36

4.3.4 Canister sampling

Canister sampling can be used for volatile hydrocarbons up to C9. Air samples are
collected in stainless steel canisters by the aid of a pump or just by opening the
valve of an evacuated canister. The canisters are sent to the laboratory for analysis
and then cleaned by evacuating it (vacuum).

4.3.5 Adsorbent tubes

Adsorbent tubes can be used for sampling of a wide number of volatile organic
compounds. The tubes can be filled with different kinds of adsorbents, depending
of which components of interest. When used as a passive sampler, there is no need
for any extra equipment. To decrease the minimum sampling period or to improve
the detection limit, the tube can be connected to a pump.

Adsorbent tubes are not suitable for some of the most volatile hydrocarbons.

4.3.6 High volume PUF-sampler

Air inlet

1-++--- PUF-plug 1
Flow Pump
meter
,-++--- PUF-plug 2

Glass
cylinder

t
Air outlet

Figure 11: The NILU high volume PUF (polyurethane foam) sampler.

The high volume PDF-sampler can be used for sampling of a wide spectre of
organic pollutants like polyaromatic hydrocarbons (PAH), dioxins, pesticides (like
DDT) etc.

The sampler consists of a glass cylinder and a filter holder. The glass cylinder
holds two polyurethane foam (PUF) plugs for trapping the gas phase of the
pollutants. The filter holder in front holds a glass fibre filter to collect pollutants
condensed on particles.

NILU TR 11/97
 37

3
The air is drawn through the sampler by a pump. 500 m of air would be a typical
sample volume for a 24 hour sample.

4.3. 7 Precipitation dust fall collection

Precipitation samples are collected in plastic cans. To avoid evaporation during
the hot season, the liquid is normally collected through a narrow inlet into a jar.
Dust fall is collected in open buckets (see Figure 12). The collection periods vary
from 1 day/week (for precipitation) to 30 days for dust fall.

Dustfall and Rain collector

snow collector NILU

Figure 12: The NILU dust fall and precipitation collector.

When analysing heavy metals, the cans are sent to the laboratory where the
samples are analysed and the cans are cleaned with acid. If no heavy metals are
analysed, only a portion of the samples are taken out of the can and sent to the
laboratory. The can is then flushed with cleaned water and used again. All
precipitation samples are stored in a cool place.

4.3.8 Semi-automatic sequential samplers

The determination of pollutant concentrations undertaken by samplers requires
that a sample is brought to the chemical laboratory for analysis.

NILU TR 11/97
38

Traditionally, sampling and analysis have been described as separate events. This
is due to the fact that until the early seventies, ambient air quality was conducted
by sampling systems that were for the most part intermittent, and which provided
average rather than real-time measurements. Intermittent sampling systems collect
gases in a solution or particles on a filter, typically over a period of 24 hours. For
most programmes of this type such a sample is collected only once every 6 day.

A few semi-automatic sequential samplers have been developed and are still
available on the marked. These have been widely used, especially in Europe, for
daily average SO 2, NO2, and PM/Black Smoke (BS) sampling.

After collection, the sample is removed from the collection device and transported
to the laboratory where it is analyzed manually by chemical or physical methods.

The air quality sampler involves four steps as shown in Figure 13; an inlet system
to bring air to a collection device where the pollution is measured or prepared for
analysis, an air flow meter where the volume of air is measured and controlled at a
constant rate and an air mover which draws air through the system.

9
Deposition
Dry and wetr----J
i eles Flow-
master
On filter:
BS, Pb, Sulfate,
TSP, elements

In liquid:
S02, N02

'- Absorpation
In bucket:
Dustfall, pH,
sulfate, nitrate,
ions,+ ...

Figure 13: A typical inlet systemfor the ambient air pollution sampler.

The inlet system must be clean and made of a material that does not react with the
air pollutants. Glass is often preferred. With long inlet systems the time required
for a volume of air to reach the collection device must be considered. In addition
to time lag, the problem of reaction between the various pollutants with each other
during the transfer may arise, due to for instance a higher temperature inside the
sampling tube than ambient air. This is particularly so when sampling nitrogen
oxides and ozone. In such cases a system with high flow rates should be
considered to reduce or minimize the time lag.

NILU TR 11/97
 39

The methods of collecting gases and particulate matter include:

adsorption
absorption
freeze-out
impingement
thermal and electrostatic precipitation
direct measurement
mechanical filtration.

The collection device is based on discrete sampling periods, semicontinuous or

continuous sampling coupled to a recorder or a computer network.

Automatic sequential samplers have been developed and used for collection of
time integrated samples with averaging times from a few hours and usually up to
24 hours.

The most commonly used device has been the bubbler, often together with a
filtration system. A chemical solution is used to stabilize the pollutant for
subsequent analysis with minimum interference by other pollutants.

To determine the pollutant concentration, it is necessary to measure the air volume

sampled. The gas flow rate or the total gas volume sampled. The gas flow rate or
the total gas volume integrated over the sampling period may be determined using
gas flow meters, rotameters, anemometers or liquid burettes. Temperature and
pressure corrections are taken to convert the air volume to standard conditions.

The air mover may be an electrical or battery-powered pump, a squeeze bulb or a

vacuum system.

NILU TR 11 /97
40

1 ®®®9

1. Intake to manifold 7. Timer

2. Manifold 8. Absorption bottle
3. Solenoid valve 9. Panel
4. Programmable sample 10. Power switch (on/off)
exchange unit 11. SHIFT, button for manual sample
5. Pump exchange 12. STOP, button for automatic stop
6. Filter holder

Figure 14: 24-hours NJLU sequential air sampler ( FK).

4.3.9 Particulate matter sampling

Particles may be collected by a variety of techniques including

• gravitational settling,
• filtration,
• electrostatic and thermostatic precipitation and
• impaction.

Gravitational settling, filtration and impaction have been the most widely used
methods for ambient particulate sampling.

The simplest particulate sampling method employs the principle of gravitational

settling of large falling particles. These are collected in a open top bucket placed
in the atmosphere for a typical period of 30 days (see Figure 12). This static or
passive sampling method require no air-moving equipment.

Hi-vol sampling
The high volume sampler has been most common in air quality monitoring
programmes world wide. The principle is shown in Figure 15.

NJLU TR 11/97
 41

Filter and
Assembly

Blower
Moter

Side View

Anderson type

Figure 15: The hi-vol samplers principle.

A collecting glass fibre filter is located upstream of a heavy-duty vacuum pump

3
which operates on a high flow rate of 1 to 2 m /min. The sampler is mounted in a
shelter with the filter parallel to the ground. The covered housing protects the
glass fibre filter from wind and debris, and from the direct impact of precipitation.
The hi-vol collects particles efficiently in the size range of 0.3-100 micrometers.
The mass concentration of total suspended particles (TSP) is expressed as ug/mr
for sampling times of usually 24 hours.

Paper tape samplers

In contrast to the hi-vol sampler, paper tape samplers are semi continuous with
averaging times of about one to two hours as normal.

Paper tape samplers draw ambient air through a cellulose tape filter. After a two
hour sampling period, the instrument automatically advances to a clean piece of
tape and begins a new sampling cycle.

Advanced paper tape samplers are equipped with densitometers for optical density
measurements during the sampling period. These instruments record changes in
light transmission which can be converted into COH (Coefficient of Haze) units.
Simpler instruments without built in densitometers necessitate manual
determination of optical density in the laboratory.

COH units are based on light transmission through the soiled filter area. The
higher the ambient particle loading, the more soiled the filter. The subsequent
increase in optical density can in most instances be directly related to mass
concentration.

In the early 70s, paper tape samplers were widely used for continuous monitoring
of particulate matter concentrations. Because of difficulties in relating data
acquired by this optical method to the gravimetric data of the hivol reference

NILU TR 11/97
42

method, most paper tape sampling has been discontinued. Paper tape samplers are
still used on a standby basis in metropolitan areas where rapid data acquisition is
essential to implement episode control plans during stagnating anticyclones.

Size Selective Samplers.

A variety of sampling devices are available that segregate collected suspended
particulate matter into discrete size ranges based on their aerodynamic diameters.
These particle samplers may employ one or more fractionating stages. The
physical principle by which particle segregation or fractionation takes place is
inertial impaction. Therefore, most such devices are called impactors.

In impactors, air is drawn through the unit and deflected from its original path of
flow. The inertia of suspended particles causes them to strike or impact a
deflecting surface, where they are collected. The size range of particles collected
on the impaction surface depends on

(1) gas velocity,

(2) particle density and shape,
(3) air flow geometry,
(4) gas viscosity, and
(5) the main free path of the gas.

Multi stage or cascade impactors can fractionate suspended particles into six or
more size fractions. In theory each stage collects particles above a certain "cut-
off" diameter which is smaller than the previous stage.

Other impactors have been developed to fractionate suspended particles into two
size fractions, i.e., coarse (from 2.5-10 µm) and fine (less than 2.5 µm). Although
these virtual or dichotomous impactors operate like a typical inertial unit, large
particles are impacted into a void rather than an impervious surface. Both size
fractions are then collected on individual membrane filter paper. A dichotomous
impactor is illustrated in Figure 16.

NlLU TR 11/97
 43

Figure 16: Size selective samplers.

a) The dichotomous impactor as used by Andersen samplers,
b) The hi-vol sampler with size selective inlet ( General Metal works)

In 1987, the primary air quality standard for particulate matter was changed from
measurements of mass particulate matter concentrations that ranged upward of
100 µm in diameter to a so-called PM10 standard, which included only those
suspended particles of less than 10 µm aerodynamic diameter.

The approved monitoring method to establish compliance with the new PM10
standard requires the use of devices that inertially separate suspended particulate
matter into one or more size fractions within the PM10 size range. A variety of
devices are likely to meet the performance specifications for the EP A reference
method for PM10, including both cascade and dichotomous impactors. Another
device, a modification of the hi-volume sampler, is also likely to be used. A
modified hi-volume sampler with a size selective inlet is shown in Figure 16b.
Collection of particles in this device is based on inertial separation of PM10
particles followed by filtration.

4.3.10 Continuous automatic monitors

Methods and instruments for measuring continuous air pollutants must be
carefully selected, evaluated and standardized. Several factors must be considered:

* Specific, i.e. respond to the pollutant of interest in the presence of other

substances,
* sensitive and range from the lowest to the highest concentration expected,
* stable, i.e. remain unaltered during the sampling interval between sampling and
analysis,
* precise, accurate and representative for the true pollutant concentration in the
atmosphere where the sample is obtained,

NILU TR 11/97
44

* adequate for the sampling time required,

* reliable and feasible relative to man power resources, maintenance cost and
needs,
* zero drift and calibration (at least for a few days to ensure reliable data),
* response time short enough to record accurately rapid changes in pollution
concentration,
* ambient temperature and humidity shall not influence the concentration
measurements,
* maintenance time and cost should allow instruments to operate continuously
over long periods with minimum downtime,
* data output should be considered in relation to computer capacity or reading
and processing.

If one consider the typical air concentrations of some pollutants of interest in air
pollution studies, it is seen from Table 6 that as we go from background to urban
atmosphere, the concentration for the most common pollutants increase roughly
by a factor 1000, in the next step from urban to emission we see another factor of
about 1000.

Table 6: Typical concentrations of pollutants in samples of interest in air

pollution.

Pollutant Background Urban ambient Stack effluents

co 0.1 ppm 5-10 ppm 2,000-10,000 ppm
S02 0.2 ppb 0.02-2 ppm 500- 3,500 ppm
NOx 0.2-5 ppb 0.2-1.0 ppm 1,500- 2,500 ppm
03 10 ppb 0.1-0.5 ppm -
Suspended particulates 10 µg/m3 60 µg/m3 35x106 µg/m3
Methane 1.5 ppm 1-10 ppm
Other hydrocarbons <ppm 1-100 ppb

Few techniques or instruments are capable of measuring the total range of 106
ppm. Also the ambient conditions (temperature, humidity, interfering substances
etc.) may differ greatly from ambient to emission measurements. The selection of
sampling system is thus influenced by the expected concentration level and the
surrounding conditions. We usually find that instruments, techniques and
analytical approaches are designed for application of specific concentration ranges
as represented by background levels, ambient urban air concentration levels and
typical stack emission concentrations.

NILU TR 11/97
 45

Figure 17: A typical monitoring station including gases, particles and

meteorology.

The most commonly used methods for monitoring some of the major air quality
indicators are discussed in the following:

Sulphur dioxide (S02)

S02 should be measured from the fluorescent signal generated by exciting S02
with UV light. The internal zero span selfcheck option includes a temperature
controlled permeation tube, TFE zero span valves and a zero air scrubber. The
zero check and calibration can be performed remotely from contact through a
RS232 command. Daily controls can thus be performed from the central
laboratory PC-system. The computer operating system continuously monitors all
critical operating points of the instrument to confirm proper operations.

Nitrogen oxides (NO and N02)

The principle of chemilumiscent reactions between NO and 03 will be used for
measuring NOx. NO and total NOx is being measured. N02 is estimated after
reduction of N02 by catalytic converter. N02 measurements can be made even in
areas with rapidly changing NO concentrations. A multi-tasking computer

NILU TR 11/97
46

operating system continuously monitors all critical operating points in the

instrument to confirm proper operations. A built-in data display presents trends,
averages, status, and historical information in digital or graph format.

MODEL 42C FLOW SCHEME

Flow
NO
,,........,_
Electronics Nole
Capillary
Ozonator

(NO Model

Sample O

Converter Pressure
(NO, Model Transducer

Pump

Figure 18: A typical flow scheme for a NO-NOx-monitor.

Ozone ( 03)
An ultraviolet absorption analyzer will be used for measuring the ambient
concentrations of ozone. The concentration of ozone is determined by the
attenuation of 254 nm UV light along a single fixed path cell. The ozone molecule
is a strong absorber of the 254 nm energy and thus the energy lost over the fixed
path is proportional to the ozone concentration in the atmosphere.

The remote control/ remote programming capability permits long distance

operations through a modem communication to the analyzer via a RS232 port.
Calibrations and controls can thus be performed from a central laboratory
computer.

Suspended particles ; TSP, PM10 and PM2_5

Gravimetric methods including a true micro weighing technology will be used to
measure ambient concentrations of suspended particulate matter. We will suggest
an instrument named "Tapered Element Oscillating Microbalance (TEOM)".
Using a choice of sampling inlets, the hardware can be configured to measure
TSP, PM10 or PM2,5. International standards and requirements recommend that at
least PM10 should be measured.

The microprocessor based unit easily accommodates all siting requirements and
provides internal data storage and advanced analogue and serial data input/output
capabilities. The technique allows for near continuous measurements and data
transfer via modem and telecommunication to a central laboratory .

NILU TR 11/97
 47

Carbon monoxide (CO)

The CO analyzer proposed for this project is a non-dispersive infrared photometer
that uses gas filter correlation technology to measure low concentrations of CO
accurately and reliable by use of state-of-the-art optical and electronic technology.
When environmental conditions change, the instrument automatically zeros itself
by drawing in air, which passed through a catalyst to remove CO, and stores the
information in the memory of a micro processor. All readings are corrected before
output.
The display, graphics, printouts and RS232 outputs are autoranging. All data can
be retrieved through a modem and telephone line to a central laboratory .

Hydrocarbons and VOC

Hydrocarbons (NMHC, Methane and THC) should be measured using a flame
ionization detector (FID).

4.3.11 Open path measurements based on optical absorption

A new generation of instruments have been developed that is based upon the prin-
ciple of differential optical absorption spectroscopy (DOAS).These are automated
methods (analyzers) that can measure pollutant concentrations in the ambient air
over a long, open path up to one kilometre in length. The system is presently de-
signed and configured for measuring several gases at the same time. The DOAS
system is based upon Beer-Lamberts absorption law. It states the relationship
between the quantity of light absorbed and the number of molecules in the light
path. Because every type of molecule, every gas has its own unique absorption
spectrum properties, or fingerprint, it is possible to identify and determine the
concentrations of several different gases in the light pass at the same time.

Distance 100 m - 10 km

Sender Reciever

h·····
Spectrogra

PC
■■■ ~D
'!dem
Printer

Figure 19: The principle of a DOAS instrument.

NlLU TR 11/97
48

The DOAS has been applied to measure path/line integrated (average)

concentrations of SO 2, NO2, ozone, toluene, benzene and some other specific
gases in the urban atmospheres. The DOAS provides a rapid, continuous
measurement of the different calibrated gases. The data can be transferred to a
central computer for treatment and presentation. The disadvantages are the
relatively high price.

4.3.12 Meteorological measurements

Meteorological data are normally collected along 10 to 100 m tall towers. In some
cases meteorological data are also needed for higher altitudes. In this case weather
balloons or aircraft measurements are used.

The surface layer data which are most important for air pollution studies and for
explaining the air quality that is being measured are most often collected along a
tower using an automatic weather station. These instruments are currently being
used in urban area investigations, for industrial air pollution studies included
impact from power plants, in most large field studies, in remote areas and in
complex terrain studies. Meteorological "surface data" such as winds, tempera-
tures, stability, radiation, turbulence and precipitation are being transferred to a
central computer via radio communication, telephone or satellite.

One of the more difficult parameter to obtain on a routine basis is the height of the
boundary layer as a function of time. This height is often related to and referred to
as the mixing height. These data have to be collected from radiosonde
observations, or they can be estimated using meteorological models including
boundary layer modelling or description.

When air quality models are being applied for concentration estimates, for
exposure modelling, for preparing information and forecasting and decision
making purposes, meteorological input data from the boundary layer are crucial.

Continuous measurement of meteorology using Automatic Weather Stations

(AWS) requires sensors for at least the most important parameters such as:

1. Wind speeds,
2. wind directions,
3. relative humidity,
4. temperatures or vertical temperature gradients,
5. net radiation,
6. wind fluctuations or turbulence,
7. atmospheric pressure.

An example of an AWS is shown in Figure 20.

NILU TR 11/97
 49

I)l)
Measurements:
Standard
• Wind Speed
• Wind Direction
• Temperature
Optional
• Relative Humldlly
• Solar Radiation
• Rain/Snow
• Standard Deviation
• Dew Point
• Soll Temperature
T • Barometric Pre11ure
• water Temperature
• Wind Run
• Delta Temperature
• Heatad Rain Gage

■JL
l

Figure 20: An Automatic Weather Station

From the continuous measurements of winds the following statistical information

should be extracted:

* Wind frequency distributions of directions (wind roses) and wind speed for
each month and for seasons, for individual stations.
* Average diurnal wind patterns (land sea breeze).
* Time evolution of winds during selected air pollution episodes
* Local wind vs. large scale (synoptic) wind (if geostrophic winds are available)
* Stability and mixing height from available information (temperature
measurements and radiosondes)
* Turbulence ( cr11/u) if available
* Frequency distribution of a selected "dispersion parameter"
* Joint frequency distributions of wind direction, wind speed and a "dispersion
parameter".

4.4 Chemical analysis

4.4.1 SO2 analysis by the use of ion chromatography
Ion chromatography is a robust and precise method. It is easy to run and
maintenance with very little sample preparation and the consumption of chemicals
are low. Overall running costs are low but the investment costs are higher than for
the other method.

NILU TR 11/97
50

4.4.2 SO2 analysis by the barium perchlorate-Thorin method

This method involves the use of many different chemicals, a cation exchange
column and a photometer. The method is labour intensive, running costs are
relatively high due to the consumption of the chemicals but investment costs are
not as high as for the ionchromatograph.

4.4.3 NO2 analysis

NO2 can be analysed by ion chromatography which is an easy method to run, but
it might cause some interference problems. The most common analysis method is
by the use of a spectrophotometric detector. This method is also easy to run and
maintenance with very little sample preparation and the consumption of chemicals
is low. Overall running costs are low and analysers are available in a wide range
of prices.

4.4.4 PM10
Particles are collected on teflon filters on the same filter holder as used for SO2
sampling (TAC-method). The filters are weighed before and after the sampling.
The weighing must be carried out in a room with constant temperature and
humidity and the filters must be conditioned by storing in the room before
weighing.

4.4.5 Lead
Lead is collected on the same filters that are used for PM10. Lead can be prepared
by boiling in sulphuric acid. An easier way to prepare the samples is by the use of
closed digestion vessels and a microwave oven. The solution is analysed by
atomic absorption. The use of a graphite oven will improve the detection level.

4.4.6 VOC analysis

VOCs are analysed in the laboratory on a gas chromatograph with a
thermodesorption unit.

A flame ionisation detector (FID) can be used for the analysis of many organic
compounds but is vulnerable for interference in complex matrixes.
A mass spectrometer detector (MS) can detect all kinds of organic compounds
with very few interference problems. The MS can also be used for identification
of unknown compounds.

NILU TR 11/97
 51

4.4. 7 Analysis of (persistent) organic pollutants

Analysis of organic pollutants (PAH, dioxins, pesticides etc.) are relatively labour
intensive and the consumption of chemicals is quite high. The procedure for
sample preparation and analysis could normally be described in five steps:

• Extraction of filters,
• removal of interferences/acid wash,
• chromatographic rinsing,
• volume reduction,
• GC/MS analysis.

For some of the pollutants it could take as much as a week from the arrival of the
sample in the laboratory until the analysis is completed.

4.4.8 Analysis ofprecipitation samples

pH and conductivity
pH and conductivity are measured directly in a portion of the sample by the use of
pH- and conductivity electrodes. The samples are checked against a number of
standards and control samples.

Analysis of anions in precipitation

Sulphate (SO/), nitrate (N03) and chloride (Cr) can all be analysed by ion
chromatography. If no ion chromatograph are available, sulphate can be analysed
spectrometric by the barium perchlorate-Thorin method, nitrate by the
spectrometric Griess method and chloride by the mercury thiocyanate-iron
method. The use of an ion chromatograph simplifies the analysis since all anions
are determined in the same run. The chromatographic method is much less labour
intensive than the other methods, the consumption of chemicals are much lower
and it gives just as good results. Overall running costs are low but the investment
costs are higher than for the other methods.

Analysis of cations in precipitation

2
Ammonium (NH/), sodium (Na'), potassium (K+), calsium (Ca +) and magnesium
2
(Mg +). can all be analysed by ion chromatography. Alternatively, the metallic
2 2
cations (Na', K+, Ca +, Mg +) can all be analysed one by one with atomic
absorption and ammonium (NH/) by the indophenol blue method. By the use of
ion chromatography, all cations can be analysed in one run. The chromatographic
method is less labour intensive than the other methods.

Analysis of heavy metals in precipitation

Heavy metals can be prepared by boiling in sulphuric acid. An easier way to
prepare the samples is by the use of closed digestion vessels and a microwave

NlLU TR 11/97
52

oven. The solution is analysed by atomic absorption. The use of a graphite oven
will improve the detection level.

4.5 Data retrieval and data handling systems

4.5.1 Data storage and transfer
At site there is a need for a data acquisition system (DAS) to receive the measure-
ment values emitted by one or several gas or dust analysers, meteorological sen-
sors or other parameters. These parameters must be stored, 5 min. and hourly ave-
rage, and transmitted to a central micro computer via modem and telephone lines.

The storage time must be several days or some weeks in case of problems with
modem, transmission lines or central computer.

The DAS-system should also consist of logic outputs remotely controlled for
external zero and span solenoid valves.

4.5.2 Software
The micro computer in the central room must be able to receive the data
transmitted by several stations equipped with monitors and DAS.

The software must manage acquisition edition and storage of the data issued by
DAS. It operates with a PC compatible micro computer. It is used particularly for
managing data from atmospheric pollution work stations, optionally linked to the
central station by phone hook-ups in the context of the atmospheric pollution
control networks.

Configuration
The configuration of the software can be totally determined by the operator
(station names, parameter names, units, adjusting scale, automatic calibration,
etc.)

Acquisition
• Acquisition of 1 minute integrated values or of average 1 hour values.
• Display of operating alarms, limit overshoots etc.
• Possibility of receiving meteorological parameters (wind speed and direction,
temperature, pressure, humidity, etc.) with dominating wind calculation.

Files facilities
• One hour (stored on harddisk) data files available for external use by the
operator

N!LU TR 11/97
 53

Edition
• Choice of display, on the micro computer screen, of the 5-minute integrated
values and average 1 hour values:
a) for all the measurements of one station
b) for one specific measurement of one station.
• Overall daily display of 1-hour average stored values, including status.
• Continuous printing every 1 min., 15 min., 30 min. or 60 min. of all
parameters.
• Printing of daily report, per station or per channel of hourly average values
including validation criteria; mini, maxi, daily average values and number of
exceeding threshold values.
• Printing of monthly report, per station of per channel of daily average values
including validation criteria; mini, maxi and average monthly values and
number of exceeding threshold values.
• Display and/or printing of daily histogram per parameter of hourly values with
validation criteria; mini, maxi and average daily values, programmed threshold
values.
• Display and/or printing of monthly histogram per parameter of daily average
values including validation criteria; mini, maxi and average monthly values,
programmed threshold values.

Calibration Operation
Cycle for zero and span remote control, programmable for each channel.
Automatic calibration for each channel on the 24h cycle basis.
Calibration report.

The data must be made available for the main user.

4.6 Quality Assurance (QA)

Data quality assurance (QA) is an important part of data acquisition and data stor-
age procedures. The data quality objectives for the monitoring network should be:

• a high data rate, sufficient to ensure acceptable temporal and seasonal

representativeness
• the data capture should be evenly distributed throughout the year, dependent
upon site characteristics and pollutants
• the data prepared for storage should be accurate, precise and consistent over
time
• the data must be traceable to accepted measurement standards.

A typical performance scheme for a complete measuring cycle applied to air

quality sampling (discontinuous/manual method) is shown in Figure 21.

Monthly data capture rates (given in percent) should be reported in the data pre-
sentation reports. The average goal should be -95% accepted data.

NILU TR 11/97
54

4. 6.1 QA at site
The need of QA undertaken at the measurement site varies with the type of
equipment used. Passive samplers need only a written protocol, while a complex
monitoring station needs protocols, calibration gas cylinders and zero air
generators. Different kinds of calibrators may also be needed to make ozone and
dilution of other gases.

Calibration line Measuring

of a Reference with a Primary Calibration
Method Standard

Calibration line Sampling Measureing

of a Reference Tasks:
Method Additional
Transport Conditions
Beginning
of a Monitoring
partial of Procedure:
Sample Processing
Calibration Control of
Errors
Development of Signals

Workout of Signals

Analytical Function

Result

Figure 21: Performance scheme for a complete measuring method.

The gas blenders should be able to dilute gases from verified high concentration
table gases to working gas level to make a multipoint calibration of monitors. The
gas blenders are also used to control the concentration of the working gas
cylinder. This is normally undertaken at a central laboratory. Rotameter to control
the air flows are needed at the site.

The air quality network sites should be routinely visited once a week by the local
site operators (LSO) and serviced every six months by equipment support units
(ESU). In case of instrument breakdown or other site problems, the LSOs have to

NILU TR I 1/97
 55

undertake non-routine site visits. The frequency of such non-routine visits provide
a useful indication of the overall smooth running of the network.

4.6.2 Network calibration

A network QA is performed as a total calibration or intercalibration, dependent
upon how the network is operated. This part of the QA system must be performed
by the central monitor laboratory or by a reference laboratory. These controls
should be undertaken regularly in 5-months or 6-months intervals. The purpose of
such (inter)calibration is to

• ensure consistency of the measurements in the network

• determine the accuracy and precision of the data
• identify deviations from standard operation procedures (SOP)
• investigate systematic measurement
• check the integrity of the site infrastructure

The tests that are undertaken include a number of performances such as

• accuracy
• response times
• noise levels
• linearity
• efficiency (of NO2 converters, HC "kickers", etc.)
• integrity of the sampling system

4.6.3 Routine controls at the reference laboratory

Well defined control routines should be developed and defined m standard
operational procedures including
• questionnaires,
• forms and schemes,
• control routine check points,

To measure air volumes the reference laboratory must also have available wet gas
meters including flow rates of 3 and 20 litres/min. A good calibrated pressure and
temperature device is also needed.

There is a need for a zero air generator which has the capability of delivering air to
gas blenders and ozone calibrators. The air must be cleaned for all components
and must be free from water vapour.

5. Meteorology
The weather on all scales in space and time acts on the transport and dilution of air
pollutants and plays different roles on the air quality that we measure and feel
(Figure 22).

NILU TR I 1/97
56

Energy
/...•······......... Time

; •··... 10 min 1 _
h _6...__h 2d 20 d 1 y 1 ~~tance

r
L..
_.. -..-----L----L-+--~'------'--+--L----t-~ 0.1 km

Figure 22: Meteorological scales in space and time.

?~~ .."

Meteorology specifies what happens to a plume ( or puff) of air pollutants from the
time it is emitted from its source until it is detected at some other location. The
motion of the air dilutes the air pollutants emitted into it. Given a known emission
rate, it is possible to calculate how much dilution occurs as a function of
meteorology or atmospheric conditions, and the resulting concentrations down-
wind of the source. This will require some basic knowledge of meteorology and
its effects on the dispersion of air pollutants.

First of all a brief introduction to the composition of the atmosphere and the
characteristics of large scale weather phenomenon will be given.

5.1 The atmosphere

The earth's surface is a boundary on the domain of the atmosphere. Transport
processes at this boundary modify the lowest 100 m to 3000 m of the atmosphere,
creating what is called the boundary layer (Figure 23). The reminder of the air in
the troposphere is loosely called the free atmosphere.

NILU TR 11/97
 57

ø (11 km)

Free Atmosphere
Troposphere
o (1 km)

Figure 23: The troposphere can be divided into two parts: a boundary layer
( shaded) near the surface and the free atmosphere above it.

The height of the troposphere varies with latitude and is highest at the Equator.
Normally only the lowest couple of kilometres are directly modified by the
underlying surface. The boundary layer can be defined as the part of the .
atmosphere that is directly influenced by the presence of the earth's surface, and
responds to surface forcing on a time scale of about an hour or less. These
forcings include frictional drag, terrain induced flow modifications, evaporation
and transpiration, heat transfer and pollutant emission.

The boundary layer thickness is quite variable in time and space, ranging from a
few tens of meters (at night time with low wind speeds and winter conditions) to
hundreds of meters to a few kilometres. Diurnal variations is one of the
characteristics of the boundary layer over land. The free atmosphere shows little
diurnal variation.

Local wind and temperature patterns play a significant role to the dilution of
pollution. The transport of pollutants emitted into the atmosphere is a function of
the local (average) wind direction. The dilution of pollution is mainly a function
of wind speed and turbulence. These factors are influenced by topography which
channels the wind, vegetation, radiation and radiation balance (stability) which is
a function of the vertical temperature profile.

The transport of the emitted air pollution is directed along the trajectory of the air
parcel in which the pollutants were emitted. The trajectory is a function of wind
direction and wind speed in the wind field. The dilution of pollutants is a function
of the atmosphere's turbulent conditions, which are presented by a 3-dimensional
variation in wind direction and wind speed. Turbulence is usually defined by
fluctuation of the wind with spatial dimensions less than the pollutant plume.

The variation of wind on all scales is the most important factor deciding the afr
pollution concentration at a receptor location. The wind observed at a certain
receptor is the sum of several effects:

• large scale wind patterns (geostrophic)

NILU TR 11/97
58

• friction (roughness change)

• thermally driven local winds
• radiation balance
• topographical features (deformation, channelling ... )

The next chapters give a short introduction to meteorology and the influence of
different meteorological factors on the transport and dilution of air pollution.

5.2 Large scale wind patterns

Wind is a result of an equilibrium produced by pressure, Coriolis and friction
forces. The pressure forces are caused directly by the existence of high and low
pressure regions in the atmosphere. In the Northern Hemisphere the air blows
counterclockwise around low pressure centres while in the Southern Hemisphere
the air blows clockwise. Weather maps show regions of high and low pressure and
also denote wind direction and wind speed (Figure 24).

P = 1010

~oo1
/7 100t,,

i Low

7
'>
Figure 24: Typical pressure pattern and associated wind field (Northern
Hemisphere).

High pressure regions are called anticyclones and these are often the source of
temperature inversions. An inversion limits the atmosphere's potentiality for
dilution of pollutant emissions.

Near the earth's surface, the friction force acts upon the wind. This force causes a
change in wind velocity and wind direction.

NILU TR 11/97
 59

5.3 Terrain induced air flow

During the diurnal circulation in mountainous regions, three-dimensional
circulations can form within and just above the valleys. A brief description of the
cross-valley-axis flow (anabatic/katabatic slope winds), the along valley-axis-flow
(mountain/valley winds), and the combined three-dimensional mountainous
circulation is presented below.

5.3.1 Wind and flow

Air flow, or wind, can be divided into three broad categories: mean wind, waves
and turbulence (Figure 25). Each can exist separately, or in combination with any
of the others. Air flow is responsible for the transport of quantities such as
moisture, heat, momentum, and pollutants. In the horizontal these transports are
dominated by the mean wind, and in the vertical by turbulence.

,:iL----M-e---
-an win~alone

Turbulence alone
2
Q-#--U-l--#---l--,IA4--1,~~1-41--.f--U---lf-,-#---l~--l---l-,l__,_~..__-

-2 t

Figure 25: Idealised picture of contributions to instantaneous wind speeds from

of (a) Mean wind alone, (b) waves alone, and (c) turbulence alone. U
is the component of wind in the x-direction.

Horizontal winds on the order of 2 to 10 mis are common in the boundary layer.
Friction causes the mean wind speed to slow down near the ground. The effect of
terrain roughness on the horizontal wind speed profile is presented in Figure 26.
Vertical mean winds are much smaller than the horizontal components, usually on
the order of millimetres to centimetres per second. Near the ground it is almost
zero. The vertical wind velocity normally increases with height up to the middle
of the boundary layer. The vertical wind velocity is very dependent on
atmospheric stability.

NILU TR 11/97
 60

600
Urban area Suburbs Level country
Gradient wind
500

400 Gradient wind

Height
(m)
300
Gradient wind

200

~
100

0
0 5 10 0 5 10 0 5 10
Wind speed (m/sec)

Figure 26: Effect of terrain roughness on the wind speed profile. The depth of the
affected layer decreases with decreasing roughness (i.e. urban area
versus suburban area).

5.3.2 Mountain and valley winds

An idealized evolution of the diurnal cross-valley circulation is shown in
Figure 27.

During night, radiative cooling of the mountain sides cool the air adjacent to the
surfaces, resulting in cold downslope or katabatic winds. These winds are
normally very shallow (2 to 20 m), and the normal velocities are within the order
of 1 to 2 m/s. Above the valley floor drainage flow is a gentle return circulation of
upward moving air that diverges toward the ridges. The chilled and heavy air
flows into the valley and collects as a cold pool. Although some of the cold air
flows down the valley axis, some can remain in the valley depending on the
topography. The resulting pool is often stably stratified throughout its depth, and
is sometimes called a valley inversion. The potential temperature profile indicate
the shallow inversion layer that started to build up in the valley bottom during the
night. The radiative cooling of the ground continued throughout the night creating
a deep cold pool throughout the valley. Pollutants emitted into this inversion can
build to high concentrations because of very slow dispersion in the vertical, and
can be hazardous to people, animals, and plant life on the slopes.

NILU TR 11/97
 61

Sunset Morning
Capping inversion Capping inversion
----------------
i
Residual
layer (RL)

I
Early night Noon
_____Capping inversion _ Capping inversion

-~',
~ ML •. RL . 'ML ~
Warm RL
.... ~/ ~ ~)'
·~'-I,;; tnv~~iof //
Cold oo ~ ~ - - ➔O

Late night z Afternoon z

A A

Capping inversion Capping inversion

----------------
--~- ' ,,, -
~
Warm RL

Cool ~ +
. Valley
i
Mixed
layer

~oldx
Y--~•-
mversion
➔Ø
I

Figure 27: Idealized evolution of the cross-valley circulation during a diurnal

cycle.

During the sunny hours after sunrise, the incoming solar radiation will warm the
mountain/valley sides and the air in contact with it faster than the air at some
distance from the slope. This differential heating sets up a circulation which is
akin to the sea breeze and is called the anabatic winds. Because of this instability
in the lower layers of air set up by the differential heating, the warm air will
stream toward and up the valley sides. The solar heating of the valley bottom and
the valley sides result in a shallow layer just above ground where temperature de-
crease with height (unstable layer). Above the shallow layer is a thin well mixed
layer and the reminder of that is left of the night time inversion layer. The depth of
the well mixed-layer increase during morning in accordance with the radiation
heating of the valley floor and -sides, with a resulting decrease of the stably strati-
fied layers. Above the valley inversion there is a gentle convergence and sub-
sidence. As this warmed air leaves the valley floor, the remaining pool of cold air
set up during night sinks to replace it. Eventually, the pool of cold air is com-
pletely eliminated and the mountainous area is now covered by warm air masses.

5.3.3 Drainage winds

At night, the cold winds flowing down the valley onto the plains are known as
mountain winds or drainage winds. Depths range from 10 to 400 m, depending on
the size and flow constrictions of the valley. Velocities of 1-5 mis have been
observed and these winds are occasionally intermittent or surging. The return
gentle circulation of warmer air aloft is called the anti-mountain wind, with

NILU TR 11/97
62

velocities of about half of the mountain wind, and depth of about twice as much
(Figure 28a).

During the day, warm air gently flowing up the valley axis is known as the valley
wind. This wind consist of a valley-floor component, and sometimes an up-incline
component along the ridge tops. The cool, slow return flow aloft is called the anti-
valley wind (Figure 28b).

Nighttime Anti-mountain
mountain and
anti-mountain 1~~;_..-.----.~:=:::=:::::::::::::~~r-j
.. ·
winds .
.... •···· .. •"•' .

•
Anti valley
Daytime top of ridge
valley and ···················· .
anti-valley ..
winds /..,··•···"············· ···valley winds

'bottom of valley

Figure 28: Along-valley winds.

( a) Night-time mountain and anti-mountain winds
(b) Daytime valley and anti-valley winds.

5.3.4 The three-dimensional circulation in mountainous regions

The combined three-dimensional picture of the mountain/valley wind system is
shown in Figure 29 and Figure 30. During night the downslope and down-valley
winds will converge just above ground, with gentle up-valley winds and
divergence aloft.

NILU TR I 1/97
 63
Night Morning Afternoon
Capping
Stable core /inversion
,-~~:;-===~ ,------,-~--~

Up floor

Figure 29: Three dimensional pictures of idealized local mountain circulation

( a) at night
(b) morning
(c) afternoon

I
~% -. . ~. ;:-.... //~
a

Night Morning
--- -o -
✓
::========================:
~I \ --------
/

~/
/ ~:,ifi{Y
' 1//,/"
-
Day
---~ Evening
'
Figure 30: Typical diurnal variations for mountain-Zvalley winds

NILU TR 11/97
64

If the terrain is such that there are converging valleys, the cooled air will converge
in the valley bottoms and accelerate downward through the main valley, with the
result that the night wind in such places may be stronger than the day breeze.

On a calm day the mountain and valley winds reveal their presence by cumulus
clouds forming over the mountains during the day and dissolving in the evening.
As in the case of the land and sea breeze, the mountain and valley winds may be
overshadowed by a general wind system.

Knowledge of topography, and hence, the mountain/valley wind system is

important for air pollution transport and diffusion in valleys, as most of the low
level sources (traffic, heating, etc.) often are located along the valley bottoms.

5.3.5 Sea and land breezes

5.3.5.1 Sea breeze
The large heat capacity of oceans and lakes reduces water-surface temperature
change to near-zero values during a diurnal cycle. The land surface, however,
warms and cools more dramatically because the small molecular conductivity and
heat capacity in soils prevents the diurnal temperature signal from propagating
rapidly away from the surface. As a result, the land is warmer than water during
the day, and cooler at night. This situation causes sea breezes.

The general feature is that during the morning there is little difference in
temperature between land and sea. During mid-morning, however, air begins to
rise over the warm land near the shoreline as a result of the solar heating from the
sun, and cooler air from the water flows in to replace it. A return circulation (the
anti-sea-breeze) aloft brings the warmer air back out to the sea where it descends
toward the sea surface to close the circulation. The depth of the sea breeze have
been observed to be on the order of 100 to 500 m, and the total circulation depth
including the return circulation can range from 500 m to 2000 m.

NILU TR 11/97
 65

Night

990mb

1000 mb

Morning

980 mb ---'-----------

990 mb ------------___:::,,)
o'?-
...- ----
1000 mb ---------- //I\"'-
Sea

980 mb C +--
II!
990 mb

1000 mb
i'\
~
Sea (cold)

Figure 31: Idealized sea or lake breeze circulation.

5.3.5.2 Land breeze.

At night, land surfaces usually cool faster than the neighbouring water bodies,
reversing the temperature gradient that was present during the day. The result is a
land breeze; cold air from land flows out to sea at low levels, warms, rises and
returns aloft towards land (anti-land-breeze) where it eventually descends to close
the circulation.

5.3.6 Deformation and separation offlow

Topographical inhomogenities often result in large spatial variations in both wind
speed and direction. The wind direction changes are most pronounced in valleys
where channelling is effective. The wind speed also changes as flow passes across
hills or obstacles. When the streamlines no longer follow the contour of the hill,
the primary flow is said to "separate" (Figure 32). This separation might cause
large turbulent eddies to develop behind the hill. If pollutants are released in the
turbulent wake zone, high ground-level concentrations may be found.

Flow separation is commonly observed on the lee side of mountains and is

especially pronounced to the lee side of sharp crests. The separated regime is

NILU TR 11/97
66

characterized by high rruxmg rates, lower velocities, and reversed eddy flow.
Figure 33 summarizes some effects of separation of the boundary layer.

Channeled flow Flow across hill

Figure 32: Deformation offlow due to channelling and airflow around and over
hills (obstacles).

- Low

- --
-pressur~
-
~ Hot
:\'-.- +--
~~----,

~
~
Bolstere,2

Plume

.~ 0

Tree-dim Neutral Stable Side-view From above

Figure 33: Various aspects offlow separation in complex terrain.

NILU TR 11/97
 67

The main aspects are that:

a) In situations where air flows from a low pressure region on a plateau to a high
pressure region at lower elevation, separation of the streamlines might occur,
sometimes with a small cavity zone close to the lee side of the mountainous
plateau with reversed circulations close to the ground.

b) For flow up a slope, heated by sunshine, separation occurs near the top.

c) In a valley with steep cliff sides, or in street canyons, cavity might occur
resulting in a very complicated flow.

d) In an actual valley with cross-valley external wind, the flow may be very
unsteady; the eddy may fill the valley at one moment and then rejoin near the
foot of the wind facing slope, gusting from time to time.

e) Eddies at the foot of the upwind side of a two-dimensional mountain

("bolsters"). In situations with strong winds and weak stability, large amplitude
lee-waves or mountain waves might form when the natural wavelength of the
air match the size of the hill. Rotor circulations near the ground under the crest
of the waves might occur causing a reverse flow at the surface under the rotors.

f) In neutral flow, the air flows around the tree-dimensional hill and up the sides
to the top, where it may be separated. The neutral flow over three-dimensional
hills is similar to that of two-dimensional hills, except for the wake structure.

g) Stratified flow over a three-dimensional hill. Below the top of the hill the flow
tend to move in horizontal planes because of the stratification. At each level the
air moves around the hill as if it were a vertical cylinder with a cross-section of
the hill at that level. This pattern breaks down in two places: over the top of the
hill, and on the lee side where the horizontal flow separates

For air pollution evaluations, the most important feature of the flow separation is
the downwind wake effect behind hills and mountains. From observations in the
atmosphere and wind tunnel studies the following general features are observed:

+ The region of separated or reversed flow may extend up to 10 hill heights

downwind (Figure 34).

+ The vertical extent of this region might be from a fraction of the hill height, H,
to as much as 2H.

+ Downwind of the separated flow region the wake region is characterized by a

deficit in the mean velocity decreasing down wind, vigorous turbulence within
the wake and an average downwind motion.

NJLU TR J 1/97
68

Undisturbed
wind profile

/Inner
__./region
.... •··

Figure 34: The cavity zone in the lee of a Gaussian ridge.

A similar separation of flow occurs for flow over a building. Figure 35 show the
flow pattern around a building. The flow separates to form a large cavity behind
the building. Reverse circulations will occur at the surface in the cavity zone.
Pollutants emitted from downwind sources in the cavity.zone will be transported
backwards and up along the building facade. Pollutants emitted in this cavity zone
tend to remain there since very poor mixing between '.ihe cavity and the main
stream occurs.

Velocity
Background profile
flow½ /
Displacement
~ Streamline
flow •·•·•·•·•···•· •····
-l---l----.....-,;••:;_•·-"c~
,.,.,•'
__ ,. .....
Wake

.
...
Building Cavity

Figure 35: Mean flow around a cubical building. The presence of a bluff
structure in otherwise open terrain will produce changes in the wind
flow generally similar to those shown here.

Figure 36 shows the evolution of pollutants emitted from an upstream source

when the separated cavity and wake behind a building interfere with the plume
dispersal. Figure 36a shows what happens when the stack plume is unobstructed

NILU TR 11/97
 69

by the cavity zone of the building but enter the wake. Downward diffusion
increases by mixing occurring in the turbulent wake. In Figure 36b the plume is
trapped in the upwind cavity zone of the building resulting in high concentrations
on the lee side of the building. It is very important that industrial plant designers
are aware of this problem and make sure that stack plumes do not interact with the
different flows set up by buildings. An empirical rule of thumb for stacks located
at or near buildings is that Hstack ~ 1.5 Hbuilding·

Figure 36: Separation effects on plume dispersion.

5.4 Turbulence
The atmosphere can disperse gases and particulate matter rapidly because it is
turbulent. Turbulent flow can be defined as having the ability to disperse
embedded gases and particles at a rapid rate. Turbulence is the primary process by
which momentum, heat, and moisture are transported into the atmosphere from
the surface of the earth and then mixed in time and space.

Turbulence can be visualized as consisting of irregular swirls of motion called

eddies. Usually turbulence consists of many different size eddies superimposed on
each other. Thus, a continuous hierarchy exists from the largest down to the
smallest eddies, with molecular diffusion occupying the bottom of the scale.

NILU TR 11/97
70

The effect of eddy motion is very important in diluting concentrations of

pollutants. An air parcel that is displaced from one level in the atmosphere to
another can carry both momentum and thermal energy with it. Obviously it will
also carry the pollution emitted into the air parcel. Hence, smoke and pollution
will be diffused by the turbulent motions in both the horizontal and vertical
directions.

The effect of different eddy sizes on a plume is shown in Figure 37.

Temperature lapse rate

~ .

Wind
-----+
Stable / inversion

Neutral

Unstable

,
Figure 37: ( a) Plume dispersing in afield of small eddies in a stable atmo-
sphere (inversion). The plume will move in a relatively straight
line, with gradual increase of its cross section.
(b) Plume dispersing in afield of well defined large eddies (near
neutral atmospheric conditions). Turbulent eddies with typical
size less than the plume dimension will disperse the plume
effectively.
( c) Plume dispersing in a field of large and various sized eddies.
This is atypical daytime situation with unstable atmospheric
conditions. The dispersed plume will both grow and meander as
it moves downwind.

NILU TR 11/97
 71

Atmospheric turbulence depends in general on the magnitude of three factors:

mechanical effects or the roughness of the ground, horizontal and vertical wind
shear, and thermal instability. These factors are described separately in the
following chapters.

5.4.1 Mechanical induced turbulence

Frictional drag on the air flowing over irregular ground causes wind shears to
develop, which frequently become turbulent. The larger the irregularities, the
greater the mechanical turbulence (
Figure 38). The magnitude of the turbulent eddies is a function of surface
roughness, elevation above ground and wind speed. The faster -the mean wind
speed, the greater the contribution by mechanical effects at the ground, with the
contribution decreasing as height increases. Also, obstacles like trees and
buildings deflect the flow, causing turbulent wakes adjacent to and downwind of
the obstacle. These types of turbulence caused by air flowing over rough surfaces
are called mechanical turbulence.

Wind

Roughness

Figure 38: Mechanical induced turbulence ( surface roughness).

Mechanical induced turbulence is caused by wind flow over uneven and rough
surfaces. Turbulence is generated by mechanical shear forces at a rate proportional
2
to (au1az) (the wind speed profile). The wind profile gradient is dependent upon
the surface roughness and the stability of the atmosphere. The velocity profile can
he described using the power law:

where m varies between 0.12 and 0.50, depending on the atmospheric conditions.

NILU TR 11/97
72

5.4.2 Thermally induced turbulence

Convection or thermally induced turbulence is defined as predominantly vertical
atmospheric motion resulting in vertical transport and mixing of atmospheric
properties. Convective eddies or turbulence arise from hydrostatic instability as
the result of surface heating (i.e. solar heating of the ground during sunny days
causes thermals of warmer air to rise) (Figure 39). These eddies are largest and
occurs at a lower frequency than eddies produced by mechanical turbulence. Note
that convective turbulence, unlike mechanical turbulence, is indirectly related to
wind shear and strongly related to stability

Temp .
.

Figure 39: Thermally induced turbulence ( solar radiation).

5.5 Atmospheric stability

In its simplest terms, the stability of the atmosphere is its tendency to resist or
enhance vertical motion, or alternatively to suppress or augment existing
turbulence. Stability is related to both wind shear and temperature structure in the
vertical, but it is generally the latter which is used as an indicator of the condition.

The atmospheric stability, or the atmospheric dispersion conditions, can be

classified as either unstable (U), neutral (N) or stable (S). A short description of

NILU TR 11/97
 73

the three individual classes of atmospheric stability is given below and also shown
in Figure 40.

Neutral

Unstable
J)

J:C:=:J
stable
=
..__D
----

cc:
Surface inversion

Elevated inversion

Figure 40: Schematic presentation of the atmospheric stability and the

corresponding dilution of air pollutants above ground level.

+ Unstable atmospheric stability (U) is common on days with strong solar

heating and low wind speed, or when cold air is being transported over a much
warmer surface. The sun warms the underlying surface and vertical turbulent
eddies are set up causing vertical dispersion of the smoke plume. For emissions
at ground level or just above ground level, the concentrations will dissolve
quickly. For stack emissions, elevated concentrations may occur at the ground
because of the turbulent motion of the lowest level of air.

NILU TR 11/97
74

+ Neutral atmosperhic stability (N) occurs at moderate to high wind speeds,

usually connected to overcast skies. High wind speeds and good mechanical
turbulence/mixing result in good horizontal and vertical mixing of the smoke
plume.

+ Stably stratified atmosphere (Ls, S) is usually confined to clear nights and

winter situations with cooling of the ground and the lower layers of air. In a
stably stratified atmosphere the temperature increase with height, and hence,
the vertical dispersion is poor. In situations when relatively warm air from the
sea is transported over land, the lower level of air will be stably stratified. This
result in poor dispersion of the smoke plume both horizontally and vertically.
For ground level sources this situation is critical because of poor vertical dilu-
tion and hence, enhanced ground level concentrations of pollution. For stack
emissions, poor vertical dilution result in high level pollution concentrations
being transported far before it touches ground.

NILU TR 11/97
 75

6. Air Pollution Modelling

Air pollution dispersion models are used to establish relationships between
emissions of air pollutants and the ambient concentrations or air quality. The basic
input to these models are information on:

+ Emission sources, location, stacks, emission rates,

+ physical/chemical properties, topographical features,
+ meteorological data, turbulent dispersion
+ air quality data, to verify estimates

Information concerning the emission of air pollutants is essential when using

source oriented dispersion models (see chapter 6.3). In the following we will
discuss how to obtain emission data as input to models.

Meteorological data are essential for calculation of the transport and dilution of air
pollutants.

Knowledge of physical properties such as surface roughness, vegetation and

topography, and knowledge of chemical reactions, are needed to estimate physical
and chemical changes (reactions) that take place after the pollutants are released
into the atmosphere.

To verify and test the dispersion models developed or established for a defined
area, air pollution measurement data collected in this same area are needed as a
platform for comparisons.

In this chapter we will describe how to obtain input data for the emissions, and we
will also present a selection of different type of dispersion models available.

6.1 Emission estimates

Emission data are needed for all types of source oriented dispersion models. The
main input to the dispersion models are the emission rates. The emissions rates
can either be estimated or taken directly from emission measurement data. In the
following chapter we will describe some of the methods available for obtaining
this information.

Emission rates can be estimated from information on production actrvity or

production numbers. The fundamental equations for emission estimates are:

Emission rate Activity rate x Emission factor or

Emission rate Production rate x Emission factor.

where

+ activity or production rate relates the amount of fuel used or material produced
in the covered time period and is given e.g. in tonnes per year;

NILU TR 11/97
76

• emission factor indicates the amount of pollutant released per unit of activity
rate and is given e.g. in kilograms of pollutant per tonne of product.

• emission rate specifies the amount of pollutant generated per unit of time and is
given e.g. in kilograms of pollutant per year.

In calculating emission levels, the spatial coverage may relate to:

• point, area, and line sources

• administrative units at different territorial levels
• the whole country

according to the background data used.

In a national emission inventory, two types can be distinguished:

• national total inventories without any spatial resolution

• national spatial inventories on a certain grid system or relating to
administrative units of a certain territorial level.

Depending on the circumstances, sources can be treated individually or

collectively:

• Individual: Point sources such as power plants, refineries, and airports. Site-
specific activity and emission data if possible data can be recorded

• Collective: Sources comprising large numbers of small emitters, 1.e. all

industrial boilers or those of a certain size are treated as a whole.

Depending on the aims of the inventory and on resources available, analysts must
decide to what extent the individual approach is to be applied. Major advantage of
this type of procedure will be an essential enlargement of information about
spatial distribution concerning both location and amount of emission.

6.1.1 Emission from area sources

Area sources are used to describe sources where geographical distribution is not
exactly known and where emissions are small but in large numbers so that they
have a significant impact on concentrations. These kind of emissions can be from
house heating, traffic, land use etc .. These emissions are normally connected to
use of different fuels in an area that is distributed according to population
distribution. It could also be used to model emissions of ammonium from
agriculture. The area sources in a city have local influence, they are linked to
consumption and emission factors are needed

NILU TR 11/97
 77

6.1.2 Emission from stationary point sources

Activity data should be linked to the emission generation process as closely as
possible. Two examples can be given

+ for emission from power plant combustion of certain fuels; (1) fuel input
instead of electricity output should be used, and (2) energy units instead of
mass units should be used. Consequently, determination of appropriate heat
values of fuels may be necessary where fuel data are available in mass units
only;

• for combustion related emissions in general: emission characteristics vary from

fuel to fuel and hence activities should be reported in this way, instead of using
a total energy approach.

One must pay special attention where both combustion and fuels and processing
of materials may have effects on emissions. Fuel mixture as well as specific
energy demands may change over time. As a consequence, both fuel input and
product output need to be accounted.

Whenever point sources are estimated individually, the estimated sum of the
activity represented by these sources should be subtracted from the estimated
collective activity. This is to avoid double-counting the individually considered
point sources when estimating the rest of the source activity emissions (the
collective approach).

As in the case of point sources treated individually in the accounting for processes
with combustion, attention should be paid to avoid double-counting of energy
consumption statistics. Reference activity data may be available from public and
private statistics, institutions or research projects. Information on fuels should
include non-commercial fuels and wastes used for energy generation.

6.1.2.1 Emission factors - point sources

In most cases emission factors from literature are not fully described.
Consequently, the user should thoroughly check whether the conditions under
which such factors have been established are well understood. The following
questions can be addressed:

+ What range of boiler size is represented?

+ Is refinery throughput referred to in terms of crude oil or total oil?
• Does the refinery source include gasoline dispatch or not?
+ Regarding process with combustion referred to in terms of material, are
combustion related emission included or not?
• Is the emission factor controlled or uncontrolled?

In deciding whether to use emission factors from an outside reference for a given
country, one must check whether comparable conditions exist, e.g. regarding raw
material characteristics, type of process, or operating conditions. Application of

NILU TR 11/97
78

per capita coefficients cannot be recommended because such parameters reflect

very specific socio-economic conditions.

As an example the estimate of SOremissions using available emission factors is

dependent upon several conditions. Parameters influencing the SOx-emission
factors might be:

• sulphur content of the fuel;

• sulphur retention in ashes;
• control efficiency, free gas desulphurization;
• type of processes.

Examples of emission rates of SO 2 from various processes are roughly:

Burning of coal lb SOifton = 38 x per cent sulphur by weight.

Burning of fuel oil lb SOif 1000 gal = 159 x per cent sulphur by weight.
Diesel engine exhaust lb SOi/1000 gal = 40, based on 0.3% S in oil.
Sulphuric acid manufacture 20-70 lb SOifton of 100% ACID.
Copper smeltinga) 1250 lb SOifton of concentrated ore.
Zinc smelting» 530 lb SO2/ton of concentrated ore.
Sulphite paper makingo 40 lb SOifton of air-dried pulp.
Coke drying 0.25 lb (SO2 + SO3)/ton of product.
a) These are for primary smelting processes.
b) A small amount of this tonnage is converted to sulphuric before discharge to acid mist the
atmosphere.
c) Assumes 90% recovery of S02.

Emission rates from a medium sized power plant boiler are given in Table 7.

Table 7: Typical emission rates from medium sized power plants using coal,
fuel oil or gas.

Pollutant Emission rate Coal Oil Gas

SO2 mg S/MJ ~400 240 <1
NOx mg NO2 /MJ 250 170 60
Particles mg/MJ 1 o*) 5 <1
As µg/MJ 1.5 0.4 -
Cd µg/MJ 0.1 0.2 <0.04
Hg µg/MJ 1.0 0.06 <0.004
V µg/MJ 7 260 <0.0003
CO2 g/MJ 110 85 57
*) 1 kg coal ~30 MJ (1.6% S in coal).

NlLU TR I 1/97
 79

Typical emission factors for particulate emissions from different sources are given
in Table 8.

Table 8: Typical emission factors for particulate emissions for selected

sources.

Emission source Emission factor

Natural gas combustion
Power plants 15 lb/million ft3 of gas burned
Industrial boilers 18 lb/million ft3 of gas burned
Domestic and commercial furnaces 19 lb/million ft3 of gas burned
Distillate oil combustion
Industrial and commercial furnaces 15 lb/thousand gallons of oil burned
Domestic furnaces 8 lb/thousand gallons of oil burned
Residual oil combustion
Power plants 10 lb/thousand gallons of oil burned
Industrial and commercial furnaces 23 lb/thousand gallons of oil burned
Coal combustion
Cyclone furnaces 2X (ash percent) lb/ton of coal burned
Other pulverized coal-fired furnaces 13-17X (ash percent) lb/ton of coal burned
Spreader stokers 13X (ash percent) lb/ton of coal burned
Other stokers 2-5X (ash percent) lb/ton of coal burned
Incineration
Municipal incinerator (multiple chamber) 17 lb/ton of refuse burned
Commercial incinerator (multiple chamber) 3 lb/ton of refuse burned
Commercial incinerator (single chamber) 10 lb/ton of refuse burned
Flue-fed incinerator 28 /biton of refuse burned
Domestic incinerator (gas fired) 15 lb/ton of refuse burned
Open burning of municipal refuse 16 lb/ton of refuse burned
Motor vehicles
Gasoline-powered engines 12 lb/thousand gallons of gasoline burned
Diesel-powered engines 110 lb/thousand gallons of diesel fuel burned
Grey iron cupola furnaces 17.4 lb/ton of metal charged
Cement manufacturing 38 lb/barrel of cement produced
Kraft pulp mills
Smelt tank 20 lb/ton of dried pulp produced
Lime kiln 94 lb/ton of dried pulp produced
Recovery furnacese) 150 lb/ton of dried pulp produced
Sulphuric acid manufacturing 0.3-7.5 lb acid mist/ton of acid produced
Steel manufacturing
Open-heart furnaces 1 .5-20 /b/ton of steel produced
Electric-arc furnaces 15 lb/ton of metal charged
Aircraft, 4-engine jet 7.4 lb/flight
Food and agricultural
Coffee roasting, direct fired 7.6 lb/ton of green coffee beans
Cotton ginning and incinerator of trash 11.7 lb/bale of cotton
Feed and grain mills 6 lb/ton of product
Secondary metal industry
Aluminum smelting, chlorination-lancing 1000 /b/ton of chlorine
Brass and bronze smelting, reverberatory
furnace 26.3 lb/ton of metal charged
a) With primary stack gas scrubber.

Sources: Air Quality Criteria for Particulate Matter, AP-49, National Air Pollution Control Administration, January
1969; and Control Techniques for Particulate Air Pollutants, AP-51, National Air Pollution Control Administration,
January 1969.

NJLU TR 11/97
80

6.1.3 Emissions from road traffic

6.1.3.1 Methodology
The emissions of CO and NOx from traffic is calculated by multiplying the traffic
intensity (cars/hour) with the length of the road (km) and an "emission factor"
(g/(km*car)). For CO2, the emission factor (grams of emission per unit fuel
consumption) is multiplied with the fuel consumption (kg/km).

The emission for a given road is a function of:

+ speed
+ road gradient
+ year of calculation (this determines the technology level of the vehicle)
+ number of cars in each vehicle class.

The emissions increase with the age of the car. There are also increased emissions
from cars in cold start mode. Both of these factors can be accounted for in a
model.

The total emission from the road network (tonnes/year) is estimated from the
mean daily traffic parameters. The peak emission calculations utilizes rush-hour
parameters.

The calculation of emissions/generation of PM10 (road dust) is usually based on a

different method than for the other components. The reason for this is that PM10
refers to a 24 hour average, whereas CO and NO2 are one hour averages.
As an example on the complexity of emission models the emission description in
the NILU model ROADAIR is included.

6.1.3.2 Vehicle classes

The vehicle fleet consists of different classes of vehicles (Table 9). The
classification is based on the relation between vehicle type and the respective
exhaust demands. The classification used in Norway is called The National
Emission Model (NU) (SFT, 1993), except that in this model it was decided not to
account for gasoline trucks, vans and buses. In 1991 those classes accounted for
approximately 2% of the total traffic work in Norway (SFT, 1993). This implies
that the two concentration calculation results will not differ significantly. Light
and heavy duty diesel trucks are for simplicity grouped together in one class. This
emission model is a part of a road traffic dispersion model called ROADAIR and
is also used at NILU in the model system KILDER that calculates long time
concentrations inn cities with complex emission fields.

NILU TR 11/97
 81

Table 9: Classification of vehicle classes in ROADAIR 3./.

Class Type Fuel Max. load Weiqht

BL1 Light cars Gasoline < 760 kg < 3.5 tonnes
DL1 Light cars Diesel < 760 kg < 3.5 tonnes
DL2 Light vans Diesel > 760 kg < 2.7 tonnes
DL3 Heavy vans Diesel > 760 kg 2. 7-3.5 tonnes
DHLL Trucks Diesel > 760 kg 3.5-10 tonnes
DHLM Trucks Diesel > 760 kg 10-20 tonnes
DHLL Trucks Diesel > 760 kg > 20 tonnes
DHB Buses Diesel > 760 ko > 3.5 tonnes

6.1.4 The conception "emission factor" for road traffic

The emission factor represents the average emission for a certain distance given in
g/km or g/kWh. ROADAIR 3.11 uses emission factors as a function of speed
(interpolation between every 10 km/h). Both the one-hour average (during rush
hour) and the 24-hour average speeds are input data to the model. The emission
factor for i.e. 60 km/h speed does not represent the emission at constant speed
fluctuation, but the emission along a road with a speed limit of 60 km/h. The
speed will fluctuate around 60 km/h, including both acceleration and retardation.
There are expected to be more accelerations and retardations for lower speeds,
(queue-driving). It is assumed a lowest average one hour speed of 10 km/h.

The following have been included when estimating emission factors for a vehicle
class:

• The vehicles within a vehicle class for a given year represent different
technology levels. The emissions from a vehicle depend on the emission
demands that were valid when the vehicle was first registered.

• The emissions from a vehicle increases with the age of the vehicle. The ageing
is a function of accumulated driving length.

+ The emissions is influenced of cold start. The effect of cold start is different for
different technology levels. It is assumed that a certain fraction of the vehicles
are in cold start mode at all times. This fraction is a function of vehicle type,
road class, area type and time of day (see chapter 3.2 in the ROADAIR
3.11, 1996).

+ Driving uphill at constant speed is equal to acceleration (concerning

emissions), and downhill is equal to retardation. Emissions from heavy duty
vehicles are approximately zero when they break, or when they drive downhill
when the gas pedal is not in use.

NILU TR I 1/97
82

6.2 The emission inventory data base

A modern environmental monitoring and information system have to handle a lot
of measured, collected and generated data. To keep track of all these data and to
make them accessible in an easy way it is convenient to organize the data in a
database system. The database system may consist of several databases which
have to serve as main storage platforms for:

+ On-line emission measurements,

+ emission and discharge data including emission modelling procedures,
+ historical data (i.e. trends) and background information (area use, population
distributions etc.)
+ procedures for consequence analysis, guideline values, use of guidelines,
regulations etc.

The data bases contain information to enable an evaluation of the actual emissions
and it include data for establishing trend analysis, warnings and to undertake
countermeasures in case of episodic high emissions.

The emission data base is an interactive platform which contains input data for
emission estimates. It contains information about sources, emission factors,
consumption data, information on location (gridded co-ordinates), stack heights,
stack parameters, fuels etc.(i.e. EMEP). The emission data base can be operated
directly by the user who can use emission models to present the emission data for
different sources. Any changes and/or additions to the emission data base will
result in updated emission estimates with links to the dispersion models and
resulting database for graphical presentation.

All emission data collected on-line will after quality assurance and quality
controls be part of larger emission data base. From this base it will be possible to
present the data graphically, and to extract data for public information purposes
etc.

The emission data can also contain information on regulations, requirements,

emission regulations. Information about regulations and plans given by local
authorities or by governmental institutions should be included in this data base, as
well as support actions and emergency procedures.

The total associated data base system can also serve as a link to a meta
information system which include information on external environmental data,
these functions might also include:

+ Navigation facilities to access the needed information

+ support for standardization activities
+ world wide web/ internet functions and other connections

NILU TR 11/97
 83

6.3 Dispersion models

Numerical and statistical models are being used in air pollution studies of various
content and complexity. The models can roughly be divided into two main types:

1. Source oriented models

2. Receptor models

The source oriented models combine information about sources ( emission

inventories), meteorology and topography to estimate concentration distributions.

Receptor models use measured concentrations of various air pollutants over long
time periods and can by statistical analyses identify source impact and the
different source's contribution to the concentration measured at specific points.
The difference in the two types of models are illustrated in
Figure 41.

Meteorological conditions
Dispersion

cB
Predict ambient
concentration
Must know:
• production/ emissions
• source strength
characteristics

Measurements
of ambient
concentrations
(air quality)
+
Source-oriented model

Must know:
• Air quality
• Aerosol chem. composition
• Many chemical species

• "Fingerprints"} in ambient air

• Variation and in emissions
Stat. models

Figure 41: Source oriented and receptor models work from different input data.

The receptor model has been applied when large data sets of good air quality data
have been available to explain the contributions from different source types. The
most commonly used type of models have, however, been the source oriented
models.

The source models estimate the atmospheres ability to transport and disperse air
pollution emissions and have been used both for gases (inert and reactive gases)
and for particles and aerosols. These models are important when evaluating the
impact of future emissions and to analyse what causes the impact on air quality in
general. These type of models have recently been linked to air quality monitoring

NILU TR 11/97
84

and surveillance programmes, and they are frequently used for impact assessment,
abatement strategy planning and for air quality planning purposes in general.

Sources 1+-----+1 Models i.---+1 Measurements

Planning
tool

Figure 42: Source oriented models establish the connection between sources and
air quality.

A large number of source oriented models are available on all scales (space and
time). The models focus on different parameters for different scales. On the
smallest scale, atmospheric turbulence, buoyancy effects, surface roughness and
fluctuations of wind speed plays an important role. On the larger scale, the large
scale weather patterns, chemical transformations and deposition are important.

Early air quality model development was based on local scale problems. Since the
1970s also long range air pollution transport models have been developed.
Advanced mesoscale dispersion models for distances 10-300 km have not been
developed especially for operational purposes. Most of the models were developed
purely for scientific reasons. Investigations of mesoscale circulations (i.e. land/sea
breeze) for input to mesoscale dispersion models have been limited.

The different types of models treat the various elements of modelling differently,
such as

• source characteristics,
• transport of pollutants,
• diffusion,
• plume buoyancy,
• deposition,
• chemical reactions etc.

NILU TR 11/97
 85

The different models may roughly be divided into the following categories:

• Gaussian plume models

• Numerical models
• Trajectory models (puff, segment, etc.)
• Box models
• Statistical models

The models may also be characterized according to the investigated pollutant

• inert passive gas,

• gases influenced by physical processes ( deposition, fall-out)
• heavy gas,
• gases subjected to chemical reactions in the atmosphere.

In the following local to mesoscale type dispersion models will be described.

These models are often applied together with monitoring and surveillance
programmes and used in air quality planning. All the models are source oriented
models.

6.3.1 The Gaussian plume model

Gaussian type models are based on Gaussian (normal) probability distribution of
the concentration (particle density) in both the vertical and horizontal direction
perpendicular to the plume centreline. These models represent simple analytical
solutions to the continuity equation which require homogenous and steady state
conditions. The model concept is shown in
Figure 43.

Plume
f
/
/
--------~---
rise I
c::=:> V
i! I Q

Stack H = h s + ~h
height

Release rate
Concentration =
Wind speed • dispersion

Figure 43: The concept of the Gaussian plume model.

NILU TR 11/97
86

Gaussian type dispersion models are the most commonly applied models in
practical use to day. The equation for calculating the concentration (C) at ground
level, assuming total reflection of the plume at the surface, can be written:

where Q = release rate (µg/s)

H = effective plume height ( h.,· + dh)
a Y, a, = dispersion parameters (m)

The co-ordinate y refers to horizontal direction perdendicular to the plume axis,

and z is the height above the ground. The ground is assumed to be flat and
uniform.

The parameters cry and cr2 are the standard deviations of the concentration
distribution in y and z directions, respectively. The parameters are usually referred
to as the diffusion parameters. The values cry and cr2 are functions of the turbulent
state of the atmosphere, which again is a function of the mechanical induced
turbulence (wind shear, wind profile) and the convective turbulence (temperature
profile).

Stability classes
In the absence of measurements to estimate cry and ø, of turbulence the turbulent
state and the stability of the boundary layer is usually divided into classes,
preferably by a simple scheme based on inexpensive measurement data. The most
widely used scheme was developed by Pasquill (1974) and was modified slightly
by Turner in 1981:

A = extremely unstable (low wind, summer, day time)

B = moderately unstable
C = slightly unstable
D = neutral (overcast high winds)
E = slightly stable
F = stable (inversions, cold winter nights)

NILU TR 11/97
 87

A simplified classification scheme has been introduced based upon temperature

gradient measurements along a meteorological tower. The following classification
can be used as input to long term average concentration estimates (CDM models):

Class Temp. gradient Correspond to:

dT (deg/100 m) Pasquill Klug Brookhaven
Unstable dT <1 A+B+C IV+V B1 + B2
Neutral -1 ~ dT <0 D 1111 + 1112 C
Slightly stable 0 ~dT <0 E II -
Stable dT ?.1 F I D

Diffusion parameters
The diffusion parameter cry and ø, can be found from empirical curves as a
function of the distance from the source (
Figure 44).

Such curves have been established by several authors (Pasquill, 1961; Gifford,
1961; Irwin, 1979) based upon various types of dispersion experiments. Today
estimates are usually performed by computers or calculators and most people
would rather have a formula than a graph.

10 000 -,-------------,-,
O"y(m)
1 000
1 000
100
100
10-
10
1.0 -+-<------~---1
0.1 1 10 100 0.1 1 10 100
Distance downwind, km Distance downwind, km

Figure 44: Dispersion coefficient cry and O"z as functions of down wind distance
from the source ( empirical values based upon dispersion
experiments).

NILU TR 11/97
88

The most widely used formula has been established as the power law of distance:

ay = ax" and a z = bx"

Numerical values for a, p, b and q have been set up for different surface
conditions, for low and high stacks, and for area sources as shown in Table 10.

Table JO: Parameter values for diffusion coefficients aY =ax", az =bx"_

Source and surface Coefficients Unst. Neutr. SI. stable Stable

specifications
Surface a 0.31' 0.22 0.24 0.27
emission p 0.89 0.80 0.69 0.59
Low stacks b 0.07 0.10 0.22 0.26
Smooth surface q 1.02 0.80 0.61 0.50
Surface and low a 1.7 0.91 1.02 -
sources (area sources p 0.72 0.73 0.65 -
Rough surface, b 0.08 0.91 1.93 -
urban q 1.2 0.70 0.47 -
High stacks a 0.36 0.32 0.31 0.31
Smooth to p 0.86 0.78 0.74 0.71
medium rough b 0.33 0.22 0.16 0.06
surface q 0.86 0.78 0.74 0.71
High stacks a 0.23 0.22 1.69 5.38
Rough surface p 0.97 0.91 0.62 0.57
b 0.16 0.40 0.16 0.40
q 1.02 0.76 0.81 0.62

The effective plume height, H

The effective plume height (H) is defined as the total plume height above the
ground, which is the sum of the stack height (h.) and plume rise (dh).

Plume rise estimates are very important when determining maximum ground level
concentrations due to emissions from stacks. The maximum ground level
concentration is roughly proportional to the inverse square of the effective stack
height. The plume rise can sometimes increase the plume height compared to the
stack height by a factor 2-10.

The plume rise is a combination of buoyancy flux

and momentum flux

M=w·V

NILU TR 11/97
 89

where: V = volume flux = w-r2

Tg = plume temperature
Ta = ambient air temperature
w vertical plume speed
r = plume radius

Assuming that F0 is the buoyancy flux at the stack exit, Briggs (1981) re-
commended for buoyancy dominated plumes (power plants, etc ... ) that

This famous "2/3 law" has shown to agree well with observations.

During stable atmospheric conditions the vertical motion of the plume ts

supressed. The plume rise is then inversely proportional to the stability

S = g(oJ'u f s, + 0.01)/ T,,

The plume rise for stable atmospheric conditions can be written:

dh = 2.6(F0 I (u · s))1'3

Ta
w

u
dh
1
)

hs
l II
II
II
I dh = 1,6 • F 113 • (10 • h8)213/u8 I

w
d
=
=
2
F = 9,81 • w • (d/2) •

exitgas velocity (m/s)

stack diameter (m)
(Tg - T.)IT.

l
Tg = plume gas temperature
I}': ~ T. = ambient air temperature
1, u. = wind speed at stack level
t
i \
I I

Figure 45: Plume rise (bent over plume).

NILU TR 11/97
90

Wind speed
The horizontal wind speed u in the Gaussian plume model can not be zero.
Anemometers measuring wind near the surface may register u = 0 ( calm
conditions). However, in the planetary boundary layer the horizontal wind speed
is very seldom calm. For modelling purposes the wind speed u is usually set to
0.5 mis for "calm conditions".

For estimating plume rise the effective wind speed at the stack height should be
used. In Gaussian plume models a simple power law formula has been applied for
this purpose:

where uIO is the observed wind at 10 m is given as a function of stability and

surface conditions (Table 11).

Table 11: The values of min the power law wind profile.

Surface Unstable Neutral Light stable Stable

Urban 0.2 0.25 0.4 0.6
Rural 0.1 0.15 0.35 0.5
"Kilder" 0.2 0.28 0.36 0.42

6.3.2 Traffic and car exhaust models

Several models have been developed to calculate air pollution from road traffic.
These models are handling many sources that vary strongly with time of day.
They therefore often use statistical models for emission calculations and combine
these with dispersion models of different types. The dispersion model is different
for street canyons and open roads (highways).

Highways
Several line source dispersion models suitable for calculating air pollution
concentrations from exhaust emissions along roads have been developed. Some of
the well known models for highways are HIWAY-2, CALINE 1-4 and GM-line.
(Petersen, 1980). NILU has chosen to use the HIWAY-2 model. The HIWAY-2
model, and a modified version of it, in which the initial dispersion due to car
induced turbulence is not a function of car speed, is utilized for roads in areas with
scattered buildings and vegetation.

Street canyons
For street canyons, the basic model used is the APRAC model (Dabberdt, W.F. et
al., 1973), a semi empirical model developed at Stanford University. In a Nordic

NILU TR 11/97
 91

co-operative study, this model was further developed based on an extensive

Swedish-Norwegian measurement data base. It has been designated " Nordic
Curbside Pollution Model", and is used extensively to calculate concentrations of
CO and NO 2 in street canyons.

In a revised version of the Nordic model, a new dispersion module for street
canyons Operational Street Pollution Model (OSPM) has been developed by the
Danish National Air Quality Laboratory. (Hertel and Berkowicz, 1989a, b). The
OSPM model describes more accurately the influence of wind direction and
height of the buildings along the street than the APRAC model does.

Integrated road and street models

Based on the APRAC/OSPM and HIW AY-2 models, NILU has developed a
model which calculates emissions and maximum concentrations for chosen air
pollution parameters along road networks (ROADAIR). (Bekkestad et al., 1996)

The ROADAIR model calculates:

• Emissions of CO, NOx and CO 2 from the traffic on each road link,
• concentrations of CO, NO 2 and PM 10 at chosen distance from the road curb for
each road link,
• road dust deposition (g/m? month) along each road link,
• population exposure to CO, NO 2 and PM 10 ,
• nuisance from air pollution experienced by persons in their residence.

Figure 46 shows a block diagram of the model, and indicate the required input
data necessary to estimate the specified output.

Figure 47 gives an overview of the necessary input data.

Figure 48 shows one example from estimates using traffic models. The relative
reduction of the maximum 1 hour concentration for different distances from the
road, as calculated by the HIW A Y module in RO ADAIR compared to measured
concentration reduction with distance. The examples show long-term concentra-
tions of black smoke (particles) and deposition of road dust (pr. m2).

The total road network modelling system can also estimate emissions and
concentrations along the whole road and street network as shown in Figure 49.

NILU TR 11/97
92

RO ADAIR - simplified block diagram of model

Input Registers
Back- Building/
Traffic Road Met.
ground pcputatlon
data data data
pollution data
I

' ' ' ' module

Dispersion '
Emissions Street Highway
Road
module canyon
model
module
dust
Nuisance
module
module
·APRAC HIWAY2

♦ +
Output
• Traffic activity • CO, NO2 concentrations, each road
• Total emissions CO, NO2, CO2 • Road dust class, each road
• Road plot, colour coded
• Population exposure, homes

Figure 46: ROADAIR. Simplified block diagram of model

Road data
(road network separated into straight road segments)
• End point positions
• Road catagory
• Position in city (centre, outskirts)
• Stepness
• Width
• Building topography (classified)
❖ Traffic data
• Annual daily traffic I max. hourly traffic
• Traffic velocity, average I during rush hour
• Heavy duty diesel traffic
❖ Building data
• Distance building-road
• No. of apartments I residents units

Figure 47: ROADAIR. Overview of necessary input data .

NILU TR I 1/97
 93

Relative
concentration Max. hourly cone.
(HIWAY, modified}
Black smoke
(measured 24 h. values)
Dust deposition
(measured monthly values)
1.0

0.5

0--+---------~--~--~--~--'
0 10 20 30
Distance from edge of road (meters)

Figure 48: Examples of reduction in air pollution concentrations with distance

from a highway, estimated by one module in ROADAIR and compared
with measurement data ..

Road network
For year 1995:
• emission estimate
• concentration
• background

Figure 49: ROADAIR. Example ofplot of road system with each road classified
according to the calculated maximum I - hourly CO concentrations.
The example shows N02 concentrations along the road system in the
city of Lillehammer.

NILU TR 11/97
94

6.3.3 Puff trajectory models

Puff trajectory models are designed to simulate dispersion from semi-
instantaneous or continuous point sources over a spatially and temporally variable
wind field. The trajectory models can normally treat multiple sources.

The plume is represented by airparcels (puffs, segments). Each air parcel

represents a segment of the plume. The dispersion of each puff can be represented
in several ways. Commonly simple Gaussian concentration distributions are
assumed along the horizontal and vertical axis. Concentrations of pollutants in
specific receptor points are estimated by the sum of a complex combination of
puffs from many sources. The puff trajectory models are often applied for
situations with complex meteorology (time and space dependent meteorology, i.e.
land/sea breeze, mountain/valley wind etc.). It is necessary to know the wind field
of the area in question.

The most commonly applied puff trajectory models represent the plume by air
parcels emitted from the stack at given time intervals. The total number of air
parcels jointly represent the plume. The puffs are transported within the wind field
and the dilution is estimated using local diffusion factors. The dispersion of the
puffs can be calculated using Gaussian concentration distributions in the puff, but
other descriptions are also possible. Different Gaussian dispersion algorithms can
be used (see ch. 6.2). The Gaussian dispersion algorithms are usually functions of
time or distance. The time dependency is most useful for puff trajectory models.

Some of the features that can be included in a puff trajectory model are

• meteorology that vary in time and space,

• multiple point sources,
• time dependent emissions,
• dry and wet deposition estimates,
• variable output averaging times,
• concentrations given in specific receptor points or in a grid.

Puff trajectory models are flexible and are valid for a number of dispersion
problems. They are however less accurate than the Gaussian dispersion models for
calculation of long time averages because it is difficult to get reliable
meteorological data, emissions and dispersion parameters on an hourly basis to
correctly describe a half year average. Calculations with puff trajectory models for
half a year also implies a lot of extra work, which is not necessary when the
Gaussian dispersion models for long term averages produce reliable results.

6.3.4 Numerical models

Numerical models are based on numerical solution of the continuity equations.
Several numerical schemes for solving the equations, varying in complexity,
accuracy and computing speed, have been applied. The solution methodology will
depend upon the scale (in space and time) of the problem. The numerical

NILU TR I 1/97
 95

transport/diffusion models overcome the difficulties of the Gaussian plume

models to simulate complicated situations like non-stationary and inhomogeneous
conditions, as well as with calm situations and weak wind.

These complicated situations require a detailed knowledge of the three-

dimensional structure and its temporal variation for the relevant meteorological
variables and, sometimes, of initial condition for the numerical solution of the
elementary equation. Thus a rather large amount of input data, additional
restrictive assumptions, or a comprehensive closed set of differential equations for
all relevant variables is necessary for an adequate treatment of the problem.

6.3.5 Box models

Box-models are based on the idea that the temporal variation of air pollution
concentrations in a clearly defined area can be described considering input and
output (sources and sinks) within a schematic "budget-box".

Box models can be divided into the following:

+ Simple box-models: Considers the whole urban area as a single box, with
spatially undifferentiated predictions of short- or long-range trends of urban air
pollution concentrations.
+ Multiple box-models: The urban area is divided into a grid of horizontal boxes.
Multiple box-models permit spatial differentiation it and estimates the fluxes in
and out of the boxes.

6.3.6 Statistical models

Statistical models associate the probability distribution of pollutant concentration
fields to a set of meteorological parameters that characterizes the actual
meteorological situation. These models commonly make use of multiple
regression analysis of measured data. The results are described by means of
empirical relations, tabulation schemes, empirical orthogonal functions, or
variational methods. Statistical models in this sense can usually not be applied in
planning or for environmental impact analyses, such as predicting the impact from
planned changes in the emission fields.

6.4 Model applications

Air quality dispersion models have been and are being used for several purposes.
Some of the most important areas in which models are of greatest importance are
m

1. siting studies,
2. for environmental planning purposes,
3. environmental impact assessment reporting.

NILU TR 11/97
96

A more detailed list of possible uses of dispersion models may contain

• calculation of stack heights for single sources,

• impact assessment from large point sources,
• estimate results of emission controls,
+ accidental release impact,
• deposition of aerosols and gases to vegetation,
+ odour evaluation,
• estimate photochemical oxidant potential,
+ impact of distant sources,
• land-use planning,
+ traffic planning,
• planning of measurement programmes,
+ analyses of measurement data,
• forecasting of episodes,
+ environmental impact assessment,
+ implementation plans.

Operational dispersion models contain the type of input data that has been
described earlier in this chapter:

• Emission data,
+ meteorology (wind, turbulence, temperature),
+ chemical reaction mechanisms,
+ deposition mechanisms.

The input to these models may come from a monitoring programme or be taken
from historical data records or pre-estimated variables.
Figure 50 indicate the procedures of an operational model.

NILU TR 11/97
 97

In the Measurements Model

Evaluations
atmosphere or estimates subroutintes

Emission Emissions Emission

Emissions i-~► -H► -H ►
inventories Q (x, y, z, t) model

}--+- tlh
Meteorology Topography Wind
Wind -H► -H ► Advection
• wind -~► y,
U, V, W (X, Z, t)
• turbulance I Turbulance I
Stability
Climatology
1 Diffusion
K (x, y, z, t)
i-H► '
Diffusion

• chemical
reactions
• deposition
~
~ •
• Transformation -H ►
Deposition
+
Chemical
Physical
• Processes

Air quality --I ►

Concentrations
(dust fall)
•
Concentration
Statistics

Figure 50: The procedure of an operational dispersion model used in practical

applications.

A dispersion model is often more useful than a measurement programme. At least

together with measured air quality data the model is superior compared to the
single point measurement data only.

The type of model to be utilized for a specific application will be dependent upon
several factors such as:

+ Accuracy
+ A vai lab le computer capacity
+ Economic resources
+ Source types (chemical compounds)
+ Point source/area source
+ Continues or puff-release
+ Terrain (type, complexity, surface)
+ Scale (time and space)
+ Averaging time for estimated concentrations

A model produces a complete picture of the concentration distribution for an area.

A source oriented model can calculate the contribution, and evaluate the
importance, of each source to the total picture.

NILU TR 11/97
98

Models can also be used to evaluate the representativity of measured data. A

screening of the maximum ground level impact from stacks and low sources can
be obtained from a simple single source Gaussian type dispersion model (Figure
51.)

Short term S02 impact

Concentration
[SO,J(l'g/m')
Max. 1-h average
Wind from east
Neutral 3 mis
4000
3000

2000

1000

500
400 t-=~~,,...--~~----~~-----4
300

200

2 3 4 5 Sv~gvlk 20 30 40 50 (km)
Nike!
Distance

Figure 51: Model estimated SO2 concentrations as a function of distance from a

Nickel smelter. About 80 % of emissions are released from 150 m tall
stacks Less than 20 % of the emissions are from low sources in the
building complex. Emission from the low sources are still shown to
dominate the ground level concentrations up to a distance of
approximately 10 km from the source ..

When models are applied together with on line measurement of meteorology and
air quality as part of the modern air quality monitoring and surveillance
programme it is also possible to establish a system for automatic air pollution
alarm.
Figure 52 present the content of alarm systems that have been established m
highly industrialized areas.

NILU TR 11/97
 99

Alarm system
for air quality

Alarm system 1-------+-------, Alarm system

for air quality for air quality

Emission
Sources
data base

Statistical
Model
optimization

Accident
Mirror
Discrepancies.____.. "Alarm"
Focus source
?

Figure 52: An on-line air pollution alarm system based upon a modern
monitoring and surveillance system.

From on-line air-quality data and meteorological data the numerical computer
models can estimate expected air pollution concentration distributions for every
selected time step (most often hourly or each 5 minute). These concentration
distributions are based upon the emission data measured or estimated for the
sources of the area. Discrepancies from single point measurements or path
integrated measurements will occur on the PC screen as an "alarm", and will tell
the user that there are high concentrations that are being measured and that should
not be expected (
Figure 52).

The air quality system will give alarm when the measured concentrations are
above certain limits. The model system can determine what causes the
concentrations, such as unfavourable meteorological conditions or accidental
releases. The authorities can further take action and find out or question the
possible source areas pointed out by the modelling system. Such a system has

NILU TR 11/97
100

been developed for two separate urban areas in an industrial region of southern
Norway.

Graphical presentation of air pollution data/concentrations of air pollution for an

area can be used to present information to the public. In Oslo, Norway, the on-line
continuous air pollution monitoring system AirQUIS has been used to issue daily
information and forecast of the air pollution levels (air quality) through local radio
stations. This information has been classified according to low, medium and high
air pollution levels. The levels of selected indicators can further be related to
national Air Quality guideline values or to international (WHO) regulations or
guidelines.

Forecasting of high air pollution episodes will have to rely upon a forecast of
meteorological conditions. A parameterization . of. the meteorological conditions
used as input to the model system can give valuable information to air pollution
episode forecasting.

The on-line surveillance and modelling system can also be used to estimate future
impact resulting from changes in the emission conditions. It can also, when
operative, be used for designing optimal abatement strategies (see chapter 8.4).

NILU TR 11/97
 101

7. Data Presentation
7.1 Air pollution statistics
Standardized statistical analysis should be performed to assess air quality trends,
changes in emissions or impact from specific types or groups of sources. The
severity of the air pollution problem or the air quality should be specified relative
to air quality guideline (AQG) values, standards or pre defined levels of
classification (e.g. good, moderate, unhealthy, hazardous )

The number of hours and days, or percentage of time when the air pollution
concentrations have exceeded AQG values should be presented. This will also
need minimum requirements of data base completeness. Long term averages
(annual or seasonal) should be presented relative to AQG. In the Norwegian
surveillance programme the winter average values of SO 2 and NO 2 are presented
on maps in percent of the national air quality guideline values.

Before undertaking statistical evaluations the data should be presented and

validated based upon a form of time series. These data must be evaluated logically
to correct for drift in instruments, and eliminate data that are identified to be
include errors. It is also important that the data are checked with other relevant
information. A an example of this is shown in Figure 53.

µg/m3 cso

40

Week -168 hour-measurements (25.05.94. 31.05.94)

Figure 53: Time plot of NOx concentrations as shown on the screen from data
quality control.

NlLU TR I 1/97
102

After an analysis of the time plot the approved data can be handled in different
ways statistically.

Air quality data are most often presented as

+ time series,
+ cumulative frequency distributions, where the frequency distribution
should be referred to air quality standards,
+ average concentration distributions at various monitoring sites as
function of wind directions (Breuer diagrams or concentration "roses"),
+ Scatter plots which can be used for interrelation between simultaneous
air quality measurements, meteorological variables or other relevant
data,
+ average concentration as function of time of day.

The statistical programmes mentioned above are the most commonly used when
evaluating measured data. The following chapters will present some examples on
how the results can be presented and used.

In addition to the measured data, statistical analysis of calculated concentrations

can give additional information of the air pollution distribution for areas where
measurement data are not available. This is usually done with the same type of
statistical methods as mentioned above.

Special statistical analysis of comparison between measured and calculated

parameters are available. Different interpolation routines are available for
handling of measured data in a grid. One such method which is frequently applied
is kriging - an interpolation of measured concentrations in a grid. Three kriging
procedures are used: simple, ordinary and universal. The presentation of spatial
distribution of background air pollution data for Europe have been based on
kriging.

Some of these statistical procedures can easily be handled in a normal spread sheet
like EXCEL on a personal computer. But some need special programs. At NILU
the AirQUIS system has been developed to take care of the data bases and some of
the statistics used for presentation of results.

Examples of concentration frequency distribution and the scatter plot are shown in
Figure 54.

NILU TR 11/97
 103

1200-.------------- 99.95
a) Nedre Strandgt. 99.9
NO, (µg/m')
Oslo 99.5 Kirkeveien, Oslo
1000 99.
98.
95.
90.
800
80.
,, 70.
~
.. ....
~ 600 C 50.
CD
E ::,
30.
i [
u.. 20.
400
10.
5.
200 2.
1.
0.5
0.1
0 0.05
0 200 400 600 800 1000 1200 10 20 30 40 50 70 200 200
observed NO, (µg/m')

Figure 54: a) Scatter plot of measured vs. observed data.

b) Cumulative frequency distribution of NO2 concentrations.

The "concentration rose" is handy when investigating the impact of specific

sources. This analyses will give the average concentration as a function of wind
direction. An example of a "concentration rose" is shown in Figure 55.

Sande Mongstad South

Fra
Mongstad
6 8 10
µg/m'

10
µg/m'
Fra
Mongstad

Figure 55: "Concentration rose", (Breuer diagram) established for two measure-
ment sites at an oil refinery.

As an example of average frequency distribution as a function of time of the day,

the occurrence (in % ) of 4 stability classes are shown in Figure 56. The following
chapters will present further examples of data presentations using various
statistical programmes.

NJLU TR 11/97
 104

Stability and wind direction

9 Summer
8 Svanvik
,,,,Neutral

-
~
~
>,
7
6 -
-
V

(.) 5
C ~ /Stable
Q)
4 - -
:::,
0- ~ ,--~
,._
Q) 3 - ~
LL
2 - - ~ - >-
~

t[t[ >- -~ -
~ I f ril.
0 - -
30 60 90 120 150 180 210 240 270 300 330 360
wind direction

Figure 56: Frequency distribution of stability as a function of wind direction,

taken from a joint frequency distribution of wind and stability.

7 .2 Emission data
Emission data are usually divided into:

• Point sources( Industrial and large domestic sources)

• Area sources (domestic heating), small scale industries, agriculture, etc.)
• Line sources (road traffic, ships etc.) ~.
~~• .,~•!
The evaluation of individual emissions or groups of emissions have~'6€en
performed to establish an overall emission survey, given as annual average
emissions. These results are often tabulated and presented as emissions from
different source categories .. Table 12 gives an example of such presentations.

Table 12: Emissions by source in Norway for the year 1992. Unit 1000 tonnes
per annum.

Sources Sulphur Nitrogen Carbon monoxide

dioxide oxides
Total 37 220 849
Mobile sources 9 176 682
Road transport 3 79 638
Water transport 4 80 6
Other mobile sources 1 17 37
Stationary combustion 8 37 121
Industrial processes 20 7 46

These results can also be presented as histograms or pie diagrams as shown in

Figure 57a.

Total emission inventories can be presented by categories and as a function of

time. The development in time, related to international agreements or local goals

NILU TR 11/97
 105

and requirements may be another way of presenting emission data as shown in

Figure 57b.

CO2 emissions in Norway in 1993 by source Total emissions 1960-1993

Coastal traffic 40 Million tonnes of CO2
and fishing

Heating

10

0
19&1 1968 197'6 1984 1993
Sou reo:SN ond SFT

Figure 57: CO2 emissions in Norway given a) by source b) as a function of time

from 1960-1994.

For emissions for specified sources where the source location is given on a
gridded map, the emission inventory can be given on a geographical information
system (GIS), as shown in Figure 58.

NILU TR 11/97
106

Emissions of SO2
1 km

-1 5

Bilbao,
Spain /
N

Figure 58: Gridded emission datafor SO, emissions in a I km X I km grid for the
Bilbao area in Spain. Unit kg/hour.

Another example of emission data presented in a grid is shown in Figure 4 in

Chapter 2.

7 .3 Meteorological data
A number of different procedures are available to handle and present measured air
quality and meteorological data statistically. The most commonly used statistical
methods for presentation of meteorological data:

• Time series of selected meteorological variables,

• wind roses (wind direction frequency distribution),
• different types of frequency distributions,
• joint frequency distribution to establish the relationships between wind
direction, wind speed, atmospheric stability and/or other variables,.
• different types of scatter plots to establish connections between different
parameters collected at the same site or at different measurement sites,
• frequency distribution of stability or other meteorological data as a function of
time of day and time of year (seasonal)

The presentation of measured meteorological data is of great importance to

understand the physical properties of the local atmospheric conditions. A
presentation of any kind of data is helpful to visualize to the user the most
important features of the data and of the meteorology and climatology of the area.
It is therefore important to choose a representative tool.

NILU TR 11/97
 107

In this chapter the attention will be drawn towards the different methods available
for presenting meteorological data. The examples shown will not cover all
possible ways of presenting results from meteorological measurements, but will
introduce the reader to presentation tools most frequent used.

7.3.1 Measurements of wind speed and wind direction

Results from wind measurements are usually presented in the form of frequency
distributions. Frequency distributions are either presented as matrixes (wind speed
versus wind direction) or as i.e. wind roses. Wind roses are used to visualize
frequency distribution of wind speed versus wind direction for different
measurement stations.

Figure 59 presents wind roses for a winter season at two sites located about
30 km apart.

Viksjøfjell Svanvik
1.10.94 - 31.3.95 1.10.94 - 31.3.95

> 6,0 m/s

4,0-6,0 mis
2,0 -4,0 mis
0,4-2,0 mis

Figure 59: Wind roses for two different measurement sites; Viksjøfjell at hill top
(low friction), Svanvik in a valley (high surface roughness) (Hagen et
al., 1996).

The wind roses shows the frequency of wind in 12 30 degree-sectors, i.e. how
often the wind blows from the different directions. The frequency distributions are
given for the following sectors: north (360°) (i.e. 360 ±15°), north-north-east
(30°), east-north-east (60°), east (90°), east-south-east (120°), south-south-east
(150°), south (180°), south-south-west (210°), west-south-west (240°), west
(270°), west-north-west (300°) and north-north-west (330°). The symbol C in the
middle of the wind rose gives the percentage of calm weather. Calm conditions
refers to hourly wind speeds less than 0.4 mis.

NILU TR 11/97
108

The wind roses in Figure 59 shows that winds from west-south-west were most
frequent at Viksjøfjell.

Winds from south and south-west were most frequent during the winter season at
Svanvik. The wind speeds were much lower at Svanvik, due to more friction at the
surface in the valley. The frequency of calm weather was 1.1 % during winter at
Viksjøfjell, and 14.7% during winter at Svanvik.

Sometimes it might be useful to compare the wind direction distribution at

different meteorological measurement stations to get an indication of the
representativeness of each of the stations for a larger area. Wind roses for the
individual measurement stations can then be presented on a map as shown below.

Figure 60 presents wind roses for 1990 from Kirkenes Airport, Viksjøfjell,
Svanvik, Nikel and Janiskoski.

c;, ; ,'
'I
I ' \\
I K '
I
I
I
I
I
Kirkenes~- ✓ '·tarpdalen ~-. ··; ~
I '- I
I \ ' 10
1 NORGE ' !~
I ' '•,c

I
I Svanvik '- 'Viksjøfjell
I
I 0 S3
I
I
S2
I ' .
I

i
I
/ ,. ...
I I
/ Kobbfoss ,
/
0 ,,"
, , , ,. __ , .'. . / / ø
/
/
/
/ r'
I
I
SI
Nikel
I
')
I
I

N
I t
,,1" Janiskoski
,.,,,, /------------------0 10 20 30
_km_......

Figure 60: Wind roses for Kirkenes Airport, Viksjøfjell, Svanvik, Nikel and
Janiskoskifor 1990 (Sivertsen et al. 1991).

NILU TR 11/97
 109

The mean wind speed may also be presented as a function of wind direction as
shown in 61. The mean wind at Ullevål in Oslo during February 1996 was 1.5
mis. The highest average speed of 2.0 mis occurred at winds from east and the
lowest average speed of 0.7 mis was with wind from north-north-west. This
indicate a channelling of winds from along north-east and around south-west. The
distribution is important for evaluation of air pollution dilution.

Wind speed (mis)

2 --------------------,
Oslo
1,8 Ullevål
1,6 -----Feb.1996
1,4
1,2
1
0,8
0,6
0,4
0,2
0
30 60 90 120 150 180 210 240 270 300 330 360
S N
Wind direction

Figure 61: Mean wind speed as a function of wind direction at Ullevål, Oslo,
February 1996.

7.3.2 Measurements of temperature

Figure 62 presents the monthly distribution of hourly averaged air temperatures at
Viksjøfjell and Svanvik (meteorological measurement stations in the northern part
of Norway) for the period l.l.1990-31.3.1991. The figure gives the maximum
temperature (highest hourly mean), mean maximum temperature, monthly mean
temperature, the SO-percentile (50% of the hourly mean values are higher and
lower than this value respectively), mean minimum temperature and minimum
temperature (lowest hourly mean).

The figure shows that the difference between summer and winter was more
pronounced at Svanvik compared to Viksjøfjell. The height above sea level and
higher mean wind speeds at Viksjøfjell explains the lesser temperature variation at
this site.

Data are missing at Viksjøfjell for the period late November to mid-January 1991
because of problems with the data logger.

NILU TR 11/97
110

40 Temperatur
Svanvik
30 1990 - 1991
u
0
20
~ 10
I!?
::i Maximum
ai 0
eai ·10
~ -20
-30 Average max.
-40
Jan Mar Mai Jul Sep Nov Jan Mar
+ Mean
40 50 percentil
Viksjøfjell
30

u
oiu,I
20 Average min.
0

'; 10
:i
ai 0

e
ai

~ -20
-10
1H t H Minimum

-30
-40
Jan Mar Mai Jul Sep Nov Jan Mar

Figure 62: Temperature statistics for Viksjøfjell and Svanvikfor every month
during the period 1.1.1990-31.3.1991 (C) (Sivertsen et al. 1991).

7.3.3 Atmospheric stability and turbulence

The thermal stability of the atmosphere is an important factor for the vertical
dilution of air pollution. The stability is measured as the vertical temperature
gradient of the atmosphere, and is also a measure of thermally induced turbulence.
The turbulence is given by the small scale fluctuations in the wind and is a
measure for the dilution of air pollutants.

The atmospheric stability in this example is measured at Viksjøfjell as the

temperature difference (~T) between 2 m a.s.l. and 10 m a.s.l. The measured
temperature differences are divided into 4 classes. Each of these 4 classes indicate
the stability of the atmosphere and hence, the vertical dilution of air pollutants.
The classes are:

Unstable ~T ~-0.5 °C
Neutral -0.5 °C < ~T ~ 0.0 ° C
Light stable 0.0 °C < ~T ~ 0.5 ° C
Stable 0.5 °C < ~T

NILU TR l 1/97
 111

Neutral atmospheric stability ( often characterized by strong winds and cloudy

conditions) and unstable atmospheric stability usually results in good dispersion
of air pollutants emitted into the atmosphere.

Summer 1990 Winter 1990-91

Stable

Light
stable

Neutral

Neutral

Unstable Unstable

4 8 12 16 20 24 4 8 12 16 20 24

Figure 63: Frequency distribution of the four stability classes at Viksjøfjell as

24 h averages during the summer and winter season. (Sivertsen et al.
1991).

During night-time and winter when there is a net outgoing radiation from the
earth, the ground cools off rapidly resulting in cold air at the surface and a
temperature increase with height (light stable /stable or inversions). An inversion
layer is formed, and the dispersion of pollutants is suppressed.

Figure 63 shows the frequency distribution of the four stability classes at

Viksjøfjell for the winter seasons 1989/90 and 1990/91 and for the summer season
1990. The figure indicates a fairly low frequency of unstable weather, mostly
during daytime and summer. Inversions, or stable atmospheric conditions,
occurred approximately 50% of the time in the winter season, more frequent
during daytime than during night. During the summer, inversions occurred in
approximately 30% of the time. At night time hours it was about 50% inversion,
at daytime only 15%. Near neutral atmospheric conditions were most frequent
during summer, noticeably during mid-day (approximately 90% of the time).

NILU TR 11/97
112

The figure also indicate that the vertical dispersion of air pollutants is better
during the summer season than during the winter season, especially during day-
time.

Table 13 shows another way of presenting the frequency distribution of the four
stability classes given for each season and averaged for one year.

Table 13: The frequency (in%) of unstable, neutral, light stable and stable
atmospheric conditions at ground level measured at Brenntangen
(Norway).

Spring Summer Fall Winter Year

Unstable 18 42 11 0 18
Neutral 46 49 50 8 38
Light stable 29 8 32 77 36
Stable 7 1 7 15 8

7.3.4 The combined wind-!stability matrix

The wind frequency distribution and the atmospheric stability may be combined in
a wind-/frequency matrix (Table 14). The matrix gives the meteorological
conditions at the specific measurement station as a function of

• four stability classes,

• four wind speed classes,
• twelve wind direction classes.

The right column of Table 14 is the basis for the wind rose presented in Figure 60.
The meteorological frequency matrix is an important input to one of the Gaussian
air pollution dispersion models utilized and produced by NILU.

NILU TR 11/97
 113

Table 14: Frequency distribution of wind and stability for 4 stability classes, 4
wind speed classes and 12 wind direction classes(%) for Viksjøfjell
1.10.1989-31.3.1990 (Sivertsen et al. 1991).

O•lh T : VIKSJØFJELL
W1nd : SVANVIK
Prra od 01.04.90. - J0.09.90.
Unal Percrnt

JOINT FREQUENCY OISTRIBUTION OF STABILITY. WINO SPEED ANO WINO OIRECTION

Cl••• I: Unat.bl• OT < - .s O•gr••• C

Cl••• li: N•utral -.s < OT < .0 Oitgru•• C
Cl&•• Ill: Light ■t.able .o < OT < .5 Degre•·• C

. ,.
Cl••• IV: Stabl• .5 < OT Orgr••• C

Wand-
. 0- 1.0 .. , •
Ca 111: u
1.0-
1 ••• or •qual

2.5 .,. •J

2 .5- 4 .0 .. ,. eve r 4.0 .. ,.

d1 r e c t i cn li 111 rv li 111 IV li 111 IV 11 li I IV Bc s e

- - - - - - - - -- - - - -- - - - - -- - - - ---- - - - -- - --- -------------------------------- - ------ -- - - - --
JO .0 1.8 1. 1 .2 .0 6.5 .2 .1 .0 2.0 .o .0 .0 .6 .0 .0 12 .6
60 .0 1. 7 .7 .1 .0 5.J .3 .0 .o 1.S .0 .0 .o .2 .0 .0 9.9
.0 1.2 .8 .1 .2 4. 1 .4 .0 .1 1.2 .0 .0 .0 .2 .0 .0 8. J
90
120 .0 .5 .4 .2 .o 1.4 .3 .0 .0 .4 .0 .0 .0 .4 .0 .0 J .8
150 .0 .9 .9 .4 .o 1. 3 .5 .1 .0 .s . .2 .0 .o .1 .0 .o 5.1
180 .0 .6 .8 .4 .o 2.5 1 .3 .3 .o 1.6 .7 .0 .0 .3 .4 .1 9.0
210 .0 .4 .6 .2 .0 2.9 1.8 .7 .0 1.8 1.3 .1' .0 1. J 1. 4 .1 12.8
240 .o .8· .8 .J .0 2. 1 1.0 .3 .0 1.8 .5 .0 .0 1.3 .J .o 9.1
270 .0 .B .4 .2 .o 1 .6 .3 .1 .o 1.4· .3 .0 .o .7 .2 .0 6.1
300 .0 1.0 .4 .1 .o 1.2 .3 .0 .0 1.5 .1 .0 .0 .7 .0 .o 5.5
330 .0 1. 1 .6 .2 .0 1. 3 .2 .0 .0 2.2 .1 .0 .0 2. J .1 .0 8. 1
360 .0 .4 .3 .1 .0 1.2 .1 .0 .0 f.O .0 .o .o .9 .0 .0 4.0
Calin .0 1.8 J. 1 .9 5.8
- - - - - - - - - - - - - - - - - - - - -- - - - --- - - - - --- - ---- --------- ------ ---- ----------- --- - ----- --- - - - -- - - - ----
Total .o 1 J. 1 10.8 J. 4 .3 31. J 6.9 1. 7 .2 17.0 J. 3 .3 .0 9.0 2.5 .3 100.0

Occur !"ener
Wind apee d
27 .3 Y.
. 6 ,.,.
40.2 Y.
1.8 .. , •
20.B Y.
3.2 .. ,.
11. 7 Y.
5.1 .,, . 100.0 Y.
2.1 ,., .
Fr•quency of occurrencr of tbe 1tabillty ela••••

Ch••• Claa, li Clan 111 Cla,, IV

Occur re-ner .6 l. 70.4 Y. 23.4 Y. 5.6 Y. 100.0 1.

7.3.5 Precipitation
A graphical presentation of precipitation rates, precipitation intensity and
precipitation as a function of wind direction is useful if calculations of wet
removal depositions are to be performed. An example of such a presentations is
shown below in Figure 64.

NILU TR 11/97
114

Precipitation Slagentangen
18
~
16
.........

-
?f2. 14
u
(1) 12
'""'"

-
s...
c.. .
0
>.
10
'
.
(.)
C 8 r,- ~
(1)
:J ~ ~
Ci
(1) 6
s...
LL
4 ,.,... =
,.,,.,,
,' ,"""
_,
2

0
ri
30 60 90 120 150 180 210 240 270 300 330 360
Wind direction

Figure 64: Precipitation as a function of wind direction at Slagentangen, Norway

( 1961-69) (Sivertsen et al. 1990).

7.3.6 The representativity of the wind measurements

NILU has performed wind measurements at Svanvik since fall 1978. Figure 65a
presents wind frequency distributions for 12 30° sectors for the winter season
1990/91 compared to the long-term average for the winter seasons 1978-89. The
frequency distributions compare well, at the data collected during 1990/91 can be
considered representation or typical for this site.

A comparison of frequency distributions collected at two different sites is

presented in Figure 65b. If all data were clustered along the diagonal one station
would perfectly represent the other. In this case it can be seen that the wind at
Nikel has a tendency of blowing to the left of that at Svanvik. Wind from south at
Nike! give simultaneously wind from south-south-west at Svanvik.

N!LU TR I 1/97
 115

25
Winter (Oct.-Mar.) Svanvik
Winter
--
";:/2.
0
20

>- 15
(.)
C:
Q)
::::, 1990/91
0-
10
....
Q)
u.
~·
5
1978-1989

0
N E s w N
Wind direction

Wind direction at Nikel

3 6 9 12 15 18 21 24 27 30 33 36 37

Number 3 8 18 5 3 2 2 2 1 3 7 22@ 156

of obs.-....
s 6 22 6 1 1 1 1 1 1 8 23 11 144

~i
>
C
9
12
2

5 1
1 2 2

3
1

2
2

1
6 3 88
47
ca
>
-
u,
ca
C
0
.;
15
18
@
177@)
8 7

23
3
3
3
8
3

4 1
2 8135
1 18 456
(.)

~ 21 1 3913@ 16 6 2 1 7 1
=s 24
'tJ 1 6 1212@ 9 1 2 1
C
~ 27 1 9 10 15 43@ 3 2 10 145
30 1 1 5 6 16 11@ 5 4 145

33 1 2 2 2 7 7 22@ 6 8144
36 1 2 4 1 1 3 2 1 18@ 5 141

37 4 7 13 35 49 63 48 17 14 12 13 11@ 53

50 100 93275 493427 395197173 148 150147 264

Figure 65: The representativity of the wind frequency distribution: a) measured

at Svanvik during the winter 1990/91 compared to the long term mean
for the winter seasons; b) simultaneously observed wind directions at
two measurement sites (Nike[ and Svanvik) to identify differences and
representativity.

NILU TR 11/97
116

7.4 Air quality data

7.4.1 Trends, changes in time
The presentation of selected air quality indicators as a function of time is a helpful
tool in understanding time variations in emissions and dispersion conditions.
Analyzing the time variation at several measurement sites, as presented in Figure
66, shows typical seasonal differences at one site compared to another.

160 I I I

140
J
--I -
120 ~
~
100

80
I I

60

40

20

Apr.91
Maajavri
Nikel
Jan.92 Viksjøfjell
- Svanvik
Jan.93

Figure 66: The variation in time of monthly SOrconcentrations at 4

measurement sites from April 1991 to January 1993.

Some sites are typically impacted during the winter season, when the predominant
wind transports SO 2 from a smelter complex towards the measurement site. One
data set (Nikel)indicate a summer maximum. This site is located downwind from
the smelter during predominant summer wind directions.

Box plots have been used by OECD-countries as an advanced air quality trend
indicator. A typical box plot is shown in Figure 67.

NILU TR 11 /97
 117

Annual average
NO, (µg/m') Urban Residential Area
Western Europe

:: ~·· · · ~""""®" . . . . @. . . . @········ @..

20
,.....L,-'--L... +-- 90th percentile

--L---'- - 70th percentile

1988 1989 1990 1991 1992 1993

No. sites 96 100 101 101 102 87 ~-----, - Median
59 35 (Peak)

Annual max. 24 h average X - Composite average

250~------------~

=r .
NO, (µg/m'} L....---.....J - 25th percentile

:: : ~:·· · :
... ·~~.... ..... ······~·····~
,.___,...J - 10th percentile

+--- 5th percentile

50

1988 1989 1990 1991 1992 1993

No. sites 131 136 139 139 139 124

Figure 67: An OECD trend analysis presenting annual N02 data ( average and
max. 24 h average)from 1988 to 1993from up to 139 measurement
sites in Western Europe.

The box plot represents a uniform method for pollutant specific (indicator) air
quality trends reporting. It increases the comparability, it can present national or
international wide trends and represents a standardized reporting procedure.

Boxplot diagrams have been generated for several combinations of regions, site
categories and defined pollutant indicators. In cases of insufficient monitoring
sites, or unavailability of data, the establishment of trend can be difficult.

From the Norwegian Surveillance programme, operated by NILU for the

Norwegian Pollution Control Authority, a simplified trend analysis is presented in
Figure 68.

NILU TR 11/97
118

3
S02, soot, N02 (µg/m ) Lead (µg/m3)

60 1,2

50 1,0

40 0,8

30 0,6

20 SO2 (Okt-Mar) 0,4

10 0,2

0_._~--.--~--.-----,-----r------.-----.----,-~0,0
1976/77 78/79 80/81 82/83 84/85 86/87 88/89 90/91 92/93

Figure 68: Air quality trends as an average for 8 selected urban areas in Norway
( 1977-94).

Data from 8 selected cities in Norway have been used to demonstrate the long
term trend of SO 2, soot, lead and NO2 in Norway over the past 20 years as shown
in Figure 68. The figure shows the development in time of the winter average
concentrations since 1976/77. The Norwegian air quality guideline values are
specified for 6 month winter averages. Hence, data presentations often mainly
contain winter average concentrations. Studied have also been performed to look
at the differences between summer and winter averages, as shown in Figure 69.

1.2 µgim' Lead ■ Winter

SQ µg/ml
■ Winter

CJ Summer 45 II Summer
40
35
0,8
30
0.6 25

0.4

0.2

76/'T7 77n8 78179 79(80

~.~.H I .I ,1 ,I ,I ,
80181 81/82 82/83 83/11.4 84185 85186 86187 87/88 88189 89i'90 90r'91 91192
Year
~----~Lb~ll.bl111J
76177 77n8 78/79 79180 80/81 81/82 82/83 83184 84185 85186 86187 87/88 88189 8MKl 90l91 91/92 92/93 93194
Year

Figure 69: Long term trends of winter and summer average lead and S02
concentrations in Norway.

The significant reduction in SO 2 levels has been caused by a shift to lighter and
sulphur poor fuel oils and a steady change to using hydro electric power for home
heating. The reduction in lead concentrations is partly caused by the introduction
of unleaded gasoline since 1983 and lowering the lead content in all gasoline since
1980.

NILU TR I 1/97
 119

The levels of soot and suspended particles decreased due to the reduced use of
heavy fuel oil until 1983. After that time most of the suspended particles in
Norwegian cities originate from automobile traffic emissions. The traffic also
causes high concentrations of NO 2 especially during cold winter days with strong
surface inversions. NO 2 is at present, together with PM 10 , the main local air
pollution problem in Norway.

7.4.2 Peak statistics

The levels of air pollution in cities within the OECD region have been evaluated
by constructing peak statistic bar charts (OECD, 1996). The bar charts show the
range of concentrations measured by monitors within the city for defined
indicators (e.g., annual maximum I-hour nitrogen dioxide). The bar charts show
the highest, composite average and lowest values of annual maximum values for
each defined indicator as segments of a bar. For example: The bar chart for annual
maximum I-hour nitrogen dioxide shows the highest, lowest and composite
average concentration of the maximum I-hour nitrogen dioxide concentration
recorded by monitoring sites located in the city.

The principal objective of the bar charts is to enable the comparison of air quality
in large cities of Member countries, to indicate where concentrations are likely to
result in acute health effects to show country-wide, regional and OECD-wide
statistics. An example urban peak statistic bar chart is shown in Figure 70.

250~-------------------------,
Urban peak statistics
NO2 (24 h aver.) 1992

200

150 ····~····•···Y.Y.l!g.!:.9.~ ~······················•················ .. s.

100

50 \Average
Range

Figure 70: Example of an urban peak statistics bar chart taken from the OECD
study (OECD, 1996)

NJLU TR 11/97
120

7.4.3 Spatial concentration distribution

When a large number of measurement site data are available it is possible to
present a spatial concentration distribution based upon statistical averaging
procedures. Such a distribution is shown in
Figure 71 for a diurnal average NOrconcentration measured in the Helwan area in
Egypt.

300

295

290

335 340 345

Figure 71: Concentration distribution of SO2 measured as a weekly average

using simple inexpensive passive samplers.

The specific distribution presented in figure 71 was measured as a SO 2 screening

study to find the maximum impacted areas. SO 2 concentrations were measured as
weekly averages using inexpensive passive samplers. The simple screening study
indicated weekly SO 2 averages of up to 50 ug/ms downwind from major industrial
areas.

NTLU TR I 1/97
 121

The number of measuring sites available for data interpolation are normally too
few to generate a picture like the one presented in
Figure 71. However, measurements together with modelling results have
frequently been used for this purpose. several examples can be given. Figure 72
represents a combination of measurement data and model results.

SO2 Viksjøfjell 1992

■ Monthly average (µg/m 3)
D Number of hours > 350 µgim 3

40

20

Januar July

Norway
Finland
••
Russia
_e
_s_it_e_s__-_,\ 0 10 20 km

Figure 72: Seasonal model estimated average S02 concentration distribution,

and observed monthly mean and number of AQG exceedances at one
site in the area.

The last example of data presentation is based on a Geographical Information

System (GIS) platform. Concentration distributions may be based upon data from
individual measuring sites, interpolated concentration fields presented on a map or
GIS platform linked to an air pollution dispersion modelling system (see
AirQUIS).

Figure 73 presents the data from background measurements and estimates of

sulphur and nitrogen deposition in Norway.

NJLU TR 11/97
122

n Dry deposition

I Wet depositio

Svovel-S Nitrogen-N

Sulphur deposition 1992

~-~-~ 0.5 g-S/m2

0.5 - 1.0
~ > 1.0

Figure 73: Dry and wet deposition of sulphur and nitrogen in Norway

A total of 42 background stations were operated in Norway in 1993 to map the

rural and background concentrations in air and precipitation of various pollutants.
Long range transport of acidifying compound is recognized to be among the most
severe environmental problems in Norway. measurements and model estimates
show that the highest concentrations of sulphur and nitrogen components occur in
the southern and south-western part of Norway (Tørseth and Joranger, 1994).

7.4.4 Presentation of estimated concentration distributions

Results from the use of air pollution dispersion models can be presented in many
ways.

Concentrations from the use of a steady state Gaussian model was presented as a
function of distance from the source in Figure 51. It is possible to present the
results for many wind speeds and many stabilities.

The spatial concentration distribution as a result of estimates applying a three

dimensional numerical model is presented in Figure 74.

NILU TR 11/97
 123

--- ABOVE 160.0

150.0 - 160.0
140.0-150.0
130.0 - 140.0
120.0-130.0

--
D 110.0-120.0
D 100.0-110.0
w 90.0-100.0
80.0- 90.0
70.0- 80.0
m 60.0- 70.0

-
□ 50.0- 60.0
40.0- 50.0
30.0- 40.0
BELOW 30.0

Figure 74: Estimated short term ( 1 h-average) ground level concentrations of

N02 estimated for the urban area of Oslo. These estimates are
performed every hour of the year based upon on-line meteorological
data as input.

The 1 h average NO2 concentrations shown in Figure 74 may be estimated every

hour of the year if required. These types of advanced modelling results have been
used as input to exposure estimates, based on the estimates of NO2 concentrations
in selected microenvironments. Urban scale models of this type are also important
for planning purposes, impact assessment and air quality short time forecasts.

7.5 User friendly presentation

Several users of environmental data will represent different needs for information.
During the last few years developments have included more user friendly
presentations. These have been produced to meet the requirements from:

+ specialists on air pollution,

+ policy makers and
+ public.

The specialist often needs a tool that gives easy access to the data with the ability
to treat these data in different ways. The specialist also want to apply the data and
prepare his own way of presenting results graphically.

The policy makers need presentations that illustrates the conclusions that the
specialist have drawn from the information available. This is usually best done
through a graphical presentation.

The public needs information on the general state of the environment. The type of
information that is needed is more general than that of the policy maker. It often
needs to cover environmental issues that is of special concern to the public. This
could be the air quality that is expected to occur in the urban area on this specific

NILU TR 11/97
 124

day. This information could be given as a short term forecast or based upon actual
on-line data.

The information may be multimedia: texts, tables, graphs, images, sound or video
dependent on the end user. The presentations have to be designed to meet the user
needs.

The information to the policy makers should be summaries and annual reports.
These reports should contain of the work that have been done during the time
period in question and the results should be presented in tables and graphs. The
tables should be in appendixes and the graphs in the main report. The reason for
presenting the two is that in further use it is necessary to know the numbers and
these are not very easy to take out of a graph.

The public needs information that is easily available. This could be done through
leaflets, Radio forecasts of the air pollution situation in several locations. It could
also be done through video screens for pollution purposes. These can give
continuous up-to-date date information on air quality measured in the area. and
predictions of the development.

This information is usually made from data that come from individual files on a
computer. These data are processed through special computer programmes. The
output of the data is presented graphically through a graphical programme. All
this work have to be done by skilled personnel. This makes the data and
information difficult to access.

Modern system are now developing where the data are stored in databases and the
results are generated through a geographical information system. These systems
makes the access to the data easy and the treatment and presentation of data easy.
The user can deduct information and make graphs through automized procedures.
This keeps the information available up-to-date for policy makers and the public.
AirQUIS is one of these systems. AirQUIS has been based upon a GIS platform. It
is easy to apply, user friendly and is, in addition to being a presentation platform,
also a planning tool.

NILU TR 11/97
 125

8. Impact assessment

8.1 The content of the environmental impact assessment (EIA)

The purpose of an Environmental Impact Assessment (EIA) is to determine the
potential environmental, social and health effects of a proposed development. It
attempts to define and assess the physical, biological and socio-economic effects
in a form that permits a logical and rational decision to be made.

Attempts can be made to reduce potential adverse effects and impacts through the
identification of possible alternative sites and/or processes. There is no general
and universally accepted definition of the EIA. The great diversity of EIA
definitions is illustrated by the following examples:

1: Impact prediction to determine the impact on the biogeophysical environment

and on man's health and well-being as a result of legislative proposals, policies,
development programmes, projects and operational procedures.
2. Impacts and benefits of a proposed development. The assessment needs to be
communicated in terms understandable by the community and the decision-
makers. Pros and cons should be identified on the basis of criteria relevant to the
area affected.
3. A total assessment of relevant environmental and resulting social effects which
may result from the fulfilling of a defined project.
4. Establish quantitative values for selected parameters which indicate the
quality of the environment before, during and after a specified action or
establishment.
5. A systematic examination of the environmental consequences of projects,
policies, plans and programmes. Its main aim is to provide decision makers with
an account of the implications of alternative courses of action before a decision is
made.

The above definitions provide a broad indication of the different concepts of the
EIA. The EIA is normally considered to be a technical exercise. The main
objective is to provide the decision makers and the public with an account of the
implications of proposed courses of action before decisions are taken.

The results of the assessment are collected into a document known as an

environmental impact statement (EIS). The EIS includes benefits and adverse
impacts considered relevant to the project and to plans and policies under
consideration. The completed EIS is one component of information upon which
the decision maker take the his actions. Other factors such as unemployment,
energy requirements or national policies may also influence the final outcome.

The EIA should be implemented at the project planning and design stage to
improve the decision making process. It must be an integral component in the
design of a project rather than something added to the technical development of a
project. This means a continuous feedback between EIA findings, project design
and locations.

NfLU TR 11/97
126

The most important consideration of potential effects of the various air pollutants
on:

• health and well being of humans

• impact on fresh water resources
• flora and fauna
+ materials (building stock and monuments).

The time scale is of great importance. Short term acute toxicity represented by
very high concentrations over short periods of time, often linked to accidental
releases or conditions leading to air pollution episodes, act differently from long
term chronic exposure. The latter type is often connected to deposition, uptake and
intake over time. Different pollutants have to be considered on different scales in
time and space

8.2 Air Pollution Impact

8.2.1 Air pollution and human health
The assessment of the air quality in the European Community is presently being
linked to the air pollution levels and to the size of populations and ecosystems
exposed to these levels. To protect the health, the concentrations of selected
harmful air pollutants should be limited and related to given ambient air quality
standards.

Several investigations have been performed by international scientific groups to

estimate the impact to human health from various air pollutants. The exposure of
humans to air pollutants is usually a mixture of different air pollution compounds
originating from different sources. It has therefore been difficult to establish
reliable dose /response relationships from actual field data.

Air quality standards and guidelines have been established based upon air
pollution impact also to the human health and well-being. The best available
background material for evaluation of health impacts is the US- EPA criteria
documents and the air quality guidelines for Europe (WHO, 1987 and 1995). The
air quality guidelines is formulated to ensure that populations exposed to
concentrations lower than the guideline values should not inflict harmful effects.
In cases where the guideline for a pollutant is exceeded, the probability of harmful
effects will increase.

The WHO guideline values for selected pollutants are presented in ch. 3 on air
quality indicators. There are also several national standards or proposed guidelines
available related to human health.

As one example of the results presented from air pollution and health studies have
been obtained from a study on the health impact of traffic air pollution in Norway.
From more than one thousand persons followed through diaries and questionnaires
the statistical analyses indicate that various symptoms of health and well being

NILU TR 11/97
 127

were correlated to exposure to traffic pollution equivalent to NO2 levels even less
than 200 ug/m' as one hour averages. Headaches, coughing, eye irritations, throat
problems and depression were some of the symptoms asked for.

3.5

Annoying noise

3.0 Anno in smell

25

20

1.5
Fatigue

20 40 60 80 100 120 140 160 180

NO, exposure (µg/m')

Figure 75: Self reported health symptoms versus NO2 exposure

The basic information used in comparing alternative strategies in air pollution

planning is the performance of exposure estimates.

8.2.2 Exposure estimates

A simplified method for performing exposure estimates was developed for
planning purposes in Oslo. The contribution of air pollution from vehicular traffic,
home heating and industry to the population exposure was calculated based upon
data for emission, dispersion and population distributions.

The calculations were carried out in a 1 kms-grid with specific calculations for
roads with high traffic and for large point sources. Based on data for; a) pollution
advection into each km2, b) local contribution within each km2 and c)
concentrations close to streets with high traffic, estimates were made of the
cumulative spatial distribution of air pollution within each km2. These
concentrations were then used together with the population distribution to
estimate a rough exposure curve for each kmz. When added for all grids the
method became a fairly robust method for obtaining a complete picture of the
population exposure to air pollutants in Oslo. The curves were presented as the
number of people living within areas of concentrations exceeding given levels.

NILU TR 11/97
128

Number of
persons (10~
Oslo,~ orway
40

1990
~ i---
30 y------

20
16000
·······•.......
•···· ...
\ \ Env. friendly
I/scenario 2010

10
.............
I\. I
4500 ···•·..........
.... / r-----
····~-- .......
80 90 11JO 110 1' 0

Concentration NO, (µg/m~

Figure 76: Estimated N02 exposure to the population of Oslo, Norway

The Norwegian State Pollution Control Authority initiated an abatement strategy

study in Oslo to evaluate the benefit of improved air quality versus the cost of
different emission reductions. A total of 38 measures were evaluated with respect
to improvement of air quality in Oslo year 2000. The basic alternative was no
change in activity except for introduction of catalytic converters installed in new
gasoline cars from 1988.

The maximum concentration levels included in this study was representative for a
cold winter day in Oslo when the emissions are captured beneath an inversion
layer which cause high impact of air pollution. The air quality guideline values
used were; 50 µg/m3 for black smoke and 100 ug/ms for SO2 and NO2.

The basic alternative for year 2000, with no emission reduced activities included
except for catalytic converters for cars, gave that about I 84 000, I 50 000 and
12 000 persons were exposed for concentrations above air quality guidelines for
SO2, soot and NO2, respectively.

To perform a cost/benefit priority of the 38 different measures, the evaluation of

cost of each measure had to be carried out.

The number of people living in each km2 combined with the concentration
distributions are used to summarize the population exposure to 24 hour mean
episodic concentration values.

Population exposure curves for SO2, NO2, CO and particulate matter were used to
evaluate future air quality as a result of alternative emission situations.

NILU TR 11/97
 129

8.2.3 Air pollution and flora and fauna

Studies of plant damage and air pollution impact on plant growth have been
performed for several individual air pollutants and for air pollution mixtures. In
the discussion of specific air quality indicators considerations of recent scientific
results on plant damage have been considered.

Also the consideration of critical loads should be taken into account. The critical
load values is defined as a quantitative estimate of the exposure to one or more
pollutants below which significant harmful effects on specified sensitive elements
of the environment do not occur according to present knowledge.

The critical level for a given area depends strongly upon geology, vegetation,
climatology, and soil properties. It might thus be difficult to generalize. It is
possible to extrapolate maps of critical levels and loads for the fresh water system.
These maps show the deposition of acid air pollutants that the water system is able
to handle before the water biotop is damaged. It is also possible to extrapolate
maps on uptake of pollutants in plants and by surfaces. This requires a vegetation
map and a model for uptake by plants.

100 km

Classification:
■ 0,8-1,0
■ 0,6-0,8
■0 0,4-0,6
0,2-0,4
0 <0,2
0 No effect
0 No data

10° 12°

Figure 77: Exceedance of critical loads for freshwater systems in Norway

An important air pollution indicator when discussing plant damage is ozone., The
phytotoxic effects of ozone have been extensively studied. In certain sensitive
species, ozone may cause direct damage in the form of necrotic spots. Tobacco

NILU TR 11/97
130

(especially the sensitive cultivator Bell W3), spinach, beans, and clover are
examples of plants that will show characteristic tissue damage symptoms if
exposed to ozone concentrations above certain levels.

Ozone also causes invisible damage, because it interferes with the photosynthesis
assimilation of carbon dioxide in the stomata. This effect has also been
systematically studied, both in laboratory (e.g. Forberg et al., 1989) and in so
called open-top chambers, where plants can be grown and exposed to different
concentration levels of ozone under field conditions (Heck et al., 1982). These
latter experiments have shown that the crop yield losses due to ozone exposure are
considerable in both Europe and in North America. Closer examination of these
data have shown that the growth reductions are related to the Accumulated
exposure of Ozone above a certain Threshold of 40 ppb (AOT40)(Fuhrer and
Achermann, 1994).

110
AOT40
R2=0,91
90

.s::. 70
iCl
g-. 50
u
~
ai 30
ai
a:

10

5000 10000 15000 20000 25000 30000 35000 40000 45000 50000
AOT 40 (ppb-h)

Figure 78: Crop growth reductions as a function of ozone exposure expressed as

ozone exceedances of 40 ppb (80 ug/ms ) (AOT40).

The workshop further recommend that the critical level of protection of

agricultural crops and forests should be set at 5000 ppb/hours and 10 000
ppb/hours, respectively, The AOT40 values for the crops to be calculated for a 3-
month period during daylight hours only, and the AOT40 value for forests for a 6-
month period. These critical levels have been set according to laboratory- and
field experiments and a reduction on plant growth of 10%.

The impact on animals is often linked to the food chain processes. Effects of
specific toxic substances, especially some toxic heavy metals, long lived

NIUJ TR 11/97
 131

chlorinated compounds, organic compounds have however, been included in the

list of air quality indicators.

8.2.4 Air quality and atmospheric corrosion

The concern for our cultural heritage and for the general life time of buildings and
constructions have increased during the last few years. Considerations for this part
of our environment and for the cost of restoration and rebuilding, should be built
into the air quality levels when considering air pollution indicators.

As for human health the impact is usually a result of mixtures of compounds

included air pollution, climate, weathering, wind, humidity, temperature, erosion,
freezing, etc.

Dose response relationships have been established for a few specific air pollutants.
For S0 2 these data have been used in cost/ benefit analyses for sulphur- reduction
measures linked to the use of fuel oil in Europe.

For a small country like Norway estimates have shown that the annual
maintenance costs on building materials caused by air pollution is more than 300
mill. NOK (60 mill US $). Table 15 indicate the savings potential related to a
decrease in S0 2 air pollution levels in Norway.

Table 15: Atmospheric corrosion costs on the Norwegian building stock.

Costs Savings
1985 1984 1985-1994
Maintenance costs 496 198 298
Allocation costs 233 93 140
Total 728 291 438

8.3 Consequence analysis

All major man made changes will have positive and negative effects on the
environment. Before these changes are made it is important to get an overview of
the consequences of the change. There are different ways of getting this
overview. The method described here is the consequence analysis.

The objective of the consequence analysis is to bring forward information on

environmental issues and the impact on a specific project to the environment.

Phases:

1. Initial screening of problems

2. Preliminary analysis
3. Full analysis

The two last steps is the contents of the consequence analysis. The amount of
work put into the different stages and how far in the analysis it is necessary to go

NILU TR 11/97
132

can be evaluated, but the environmental consequence analysis shall contain direct
and indirect consequences for :

• natural environment
• natural resources
• future management of natural resources
• man made environments
• human health

A check list of necessary steps can be made, and NORAD and the Norwegian
authorities have presented typical check lists for consequence analysis.

An initial screening has the objective of helping project desk officers and
planners to assess a project in relation to environmental impacts. The initial
assessment shall provide a survey of environmental impacts likely to ensue if a
project is implemented. Usually an initial assessment will be based on easily
accessible information, former research, the local populations views etc ..

Only potential environmental impacts, direct and indirect, are identified in the
initial assessment. Estimates are not assumed to be substantiated by special
accounts or registrations, but rather come under full assessment .

The consequence analysis must take into consideration the following points

• technical description of the project

• the situation in the area today
• consequences for the society
+ consequences for the environment and natural resources
• emissions to air
• emissions to water
• noise
• waste management
• impact on the landscape

The contents of the consequence analysis must be made available to the public. It
is also necessary the government put forward guide lines on how to make a
consequence analysis and the issues that have to be investigated before a
permission of releases to air will be given from the government.

NlLU TR 11/97
 133

8.4 Optimal abatement strategy planning

The process of developing an Air Quality Management Strategy (AQMS), for an
urban area includes many steps. The most important of these are listed on the next
page:

* - identifying sources
* - quantifying sources emission inventory

* - monitoring of air pollution Assessment

* - assessing the exposure (impact) situation

* - identifying source - exposure relations

* - estimating the relative importance of the exposure of

various AP sources

* - assessing environmental damage

* - investigating control (abatement) options

* - performing cost-benefit or cost-effectiveness analysis Control

* - developing a control strategy and an investment plan

* - developing institutions/regulations/enforcement
* -establishing an Air Quality Information System (AQIS) Surveillance

As shown above, the AQMS consists of two main components, which are
assessment and control. In parallel with the AQMS development, and to facilitate
checking the effectiveness of the air pollution control actions, a third component
is necessary, which is surveillance.

The process of attaining acceptable urban air quality is definitely long term, and it
is dynamic. The urban area develops, and population, sources and technology
change. Throughout this process, it is very important to have an operating
Information System of Air Quality (AQIS), in order to

+ keep the authorities and the public well informed about the short-term and
long-term air quality development,
• control the results of abatement measures, and thereby,
• provide feed-back information to the abatement strategy process.

NILU TR 11/97
134

The basic concept for an Air Quality Management Strategy contains the following
main components:

• Air Quality Assessment

• Environmental Damage Assessment
• Abatement Options Assessment
• Cost Benefit Analysis or Cost Effectiveness Analysis
• Abatement Measures
• Optimum Control Strategy

The Air Quality Assessment, Environmental Damage Assessment and Abatement

Options Assessment provide input to the Cost Benefit or Cost Effectiveness
Analysis, which is also based on established Air Quality Objectives (i.e.
guidelines, standards) and Economi c Objectives (i.e. reduction of damage costs).
The final result of this analysis is Optimum Control Strategy.

The establishment and follow-up of the AQMS require that an integrated system
for continued air quality management is established/completed in Jakarta. A
system for air quality management requires continuing activities on the urban
scale in the following fields:

• Inventorying of air pollution activities and emissions

• Monitoring of air pollution and dispersion parameters
• Calculation of air pollution concentrations, by dispersion models
• Inventorying of population, materials and urban development
• Calculation of the effect of abatement/control measures
• Establishing/improving air pollution regulations.

These activities, and the institutions necessary to carry them out, constitutes the
System for Air Quality Management that is a prerequisite for establishing the
Strategy for Air Quality Management (AQMS).

In megacities in developing countries, a build-up period of several years should be

considered to establish a complete system for Air Quality Management.

During this development period, interm ediate strategies for controlling the present
air pollution problems and their development must be worked out. These
intermediate strategies must be based on existing data, and additional information
and data that can be acquired over a relatively short time (-1 year). This data base
will not be complete, but the intermediate strategies will represent the optimum
control strategy, given the data available.

NILU TR 11/97
 135

8.5 Cost/benefit analysis (example Manila)

8.5.1 Action plan
Through the work carried out in the local working groups, a large number of
proposed actions and measures has been listed, and categorized within the
following categories (Larssen et al., 1995):

• Improved fuel quality.

• Technology improvements.
• Fuel switching.
• Traffic management.
+ Transport demand management.

Each of the proposed actions were described regarding its effect (benefit), costs,
policy instruments, time-frame of instigation, and institutions responsible.

A selection of "obvious" technical measures for possible short-term introduction

was made, and cost-benefit analysis carried out for each measure separately.

The Table below gives a summary of the cost-benefit analysis. For all of the
selected measures except cleaner fuels in power plants, the calculated benefits are
very substantial, in the tens of millions of USD annually, and the benefits are, as a
rule, much higher than the estimated costs.

Table 16: Benefits and costs of selected abatement measures, annual figures.

Abatement Benefits Cost of Time frame,

Measure Avoided Reduced costs measure effect of
effects mill USD mill USD measure
Address gross polluters 160 deaths 16-20 0.08 Short-term
(Anti Smoke Belching Campaign) 4 mill RSD

Improving diesel quality, vehicles 94 deaths 10-12 10 2-5 years

2.5 mill RSD

Inspection/maintenance, vehicles 310 deaths 30-40 5.5 2-5 years

8 mill RSD

Clean vehicle standards 895 deaths 94-116 10-20 5-10 years

24 mill RSD

8.5.2 Future air quality for some abatement scenarios

In order to be able to draw overall conclusions regarding the possibility to
improve the air pollution situation in Metro Manila, two combined future
scenarios have been defined:

• "Common environmental technology scenario", based on a comprehensive

strategy to address "smoking" diesel fuelled vehicles including introduction of

NILU TR 11/97
136

clean diesel fuel; introduction of unleaded fuel and clean vehicle emissions
standards; further reduction of sulphur contents in fuel oils.
+ "A fuel shift scenario", involving gradual shift to LNG for energy production,
and introduction of LPG (and CNG) as automotive fuel for buses and trucks
particularly.

NILU TR 11/97
 137

9. References
Bekkestad, T., Torp, C.,Tønnesen, D. and Larssen, S. (1996) Programme
documentation ROADAIR version 3.11. Kjeller (NILU TR 21/96).

Briggs, G.A. (1975) Plume Rise Predictions. In: Lectures on Air Pollution and
Environmental Impact Analyses. Workshop Proceedings, Boston, Mass., Sept.
29-Oct. 3, 1975. Boston, Mass., American Meteorological Society, pp. 59-111.

Briggs, G.A. (1984) Plume Rise and Buoyancy Effects. In: Atmospheric Science
and Power Production. Darryl Randerson (Ed.). Oak Ridge, TE., Technical
Information Center, Office of Scientific and Technical Information, United
States Department of Energy (DOE Report DOEffIC-27601), pp. 327-366.

Bøhler, T. (1987) Users guide for the Gaussian type dispersion models CONCX
and CONDEP. Lillestrøm (NILU TR 8/87)

Bøhler, T. (1996) MEPDIM Version 1.0. Model description. Kjeller (NILU TR

7/96).

Dabberdt, W.F., Ludwig, F.L. and Johnson, W.B. (1973) Validation and
application of an urban diffusion model for vehicular pollutants. Atmos.
Environ., 7, 603-6 I 8.

Eggleston et al. (1991) CORINAR Working Group on Emission Factors for

Calculating 1990 Emission from Road Traffic. Volume 1. Methodology and
Emission factors. Final Report. Brussels.

Forberg, E., Aarnes, H., Nilsen, S. and Semb, A. (1987) Effect of Ozone on Net
Photosynthesis in Oat (Avena sativa) and Duckweed (Lemna gibba). Environ.
Poll., 47, 285-291.

Fuhrer, J. and Achermann, B. (1994) Critical Levels for Ozone. A UN-ECE

workshop report. Liebefeld - Bern, Swiss Federal Station for Agricultural
Chemistry (Schriftenreihe der FAC Liebefeld, No. I 6).

Gifford, P.A., Jr. (1961) Use of Routine Meteorological Observations for

Estimating Atmospheric Dispersion. Nucl. Saf, 2(4), 47-57.

Gram, F. (1996) The "Kilder" Air Pollution Modelling System. Version 2.0.
Kjeller (NILU TR 12/96).

Grønskei, K., Walker, S.E. and Gram F. (1993) Evaluation of a model for hourly
spatial concentration distributions. Atmos. Environ., 27B, 105-120.

Gryning, S.E., Holtslag, A.A.M., Irwin, J.S. and Sivertsen, B. (1987) Applied
dispersion modelling based on meteorological scaling parameters. Atmos.
Environ., 21, 79-89.

NILU TR 11/97
138

Hagen, L.O., Sivertsen, B., Johnsrud, M. and Bekkestad, T. (1996) Air Quality
Monitoring in the Border Areas of Norway and Russia. Progress Report April-
September 1995. Kjeller (NILU OR 40/96). (in Norwegian).

Hanna, S.R., Briggs, O.A. and Hosker, R.P. (1982) Handbook on atmospheric
diffusion. U.S. Springfield, Virginia, Department of Commerce, (DOE/TIC-
11223).

Heck, W.W., Taylor, O.C., Adams, R., Bingham, G., Miller, J., Preston, E. and
Weinstein, L. (1982) Assessment of crop loss from ozone. J. Air Poll. Contr.
Ass., 32, 353-361.

Irwin, J.S. (1979) Estimating Plume Dispersion - A Recommended Generalized

Scheme. In: Preprints of Fourth Symposium on Turbulence, Diffusion, and Air
Pollution. Reno, Nev., Jan. 15-18, 1979. Boston, Mass., American
Meteorological Society, pp. 62-69.

Knudsen, S. and Hellevik, 0. (1992) INPUFF 2.0. A multiple source Gaussian

Puff dispersion algorithm with NOx/SO2 chemical reactions and wet
deposition. Lillestrøm (NILU IR 3/92).

Larssen, S. and Torp, C. (1993) Documentation of RoadAir 2.0. Lillestrøm

(NILU TR 12/93).

Larssen, S. et al. (1995) URBAIR - Urban Air Quality Management Strategy in

Asia, Metro Manila City Specific Report. Kjeller (NILU OR 57 /95).

Mikkelsen, T., Nyren, K., Thykier-Nielsen, S. and Larsen, S. (1987) Rise

Mesoscale PUFFmodel, RIMPUFF, version 2.0. Roskilde (Risø-M-2673).

OECD (1994) Environmental Indicators. Paris, Organisation for Economic co-

operation and Development.

Pasquill, F. ( 1961) The Estimation of the Dispersion of Windborne Material.

Meteorol. Mag., 90, 33-49.

Pasquill, F. (1974) Atmospheric Diffusion, 2nd ed. New York, John Wiley &
Sons.

Paumier, J., Sten sen, D., Kelly, T., Bollinger, C. and Irwin, J.S. (1986) MPDA-1:
A meteorological processor for diffusion analysis, User's Guide. Research
Triangle Park N.C, U.S. Environmental Protection Agency (EPA-600/8-
86/011) (NTIS PB 86-171 402/AS).

Petersen, W.B. (1980) User's guide for HIW AY-2. A highway air pollution
model. Research Triangle Park, N.C., Environmental Protection Agency (EPA-
600/8-80-018).

NILU TR 11/97
 139

SFT (1993) Utslipp fra veitrafikken i Norge. Dokumentasjon av

beregningsmetode, data og resultater. Oslo (SFT -rapport nr. 93: 12).

Sivertsen, B. (1979) Luftkvalitetsmodeller. Sluttrapport NORD FORSK prosjektet

mesoskala spridningsmodeller. Helsingfors (NORDFORSK,
Miljovårdssekretariatet publikation 1979-1 ).

Sivertsen, B. (1980) The application of Gaussian dispersion models at NILU.

Lillestrøm (NILU TR I 1/80).

Sivertsen, B., Braathen, O.A, Larssen, S., Schjoldager, J. and Skogvold, O.F.
( 1990) Luftforurensning. En serie foredrag fra NILU. Lillestrøm (NILU TR
5/90).

Sivertsen, B., Hagen, L.O., Hellevik, 0. and Henriksen, J.F. (1991) Air quality in
the border areas of Norway and USSR (1990-91). Lillestrøm (NILU OR
69/91). (in Norwegian).

Sivertsen, B., Baklanov, A., Hagen, L.O. and Makarova, T. (1992) Air Pollution
in the Border Areas of Norway and Russia. Summary Report 1990-91.
Lillestrøm (NILU OR 8/92).

Sivertsen, B. (1994a) The use of air quality indicators in Norway. Kjeller (NILU
TR 19/94).

Sivertsen, B. (1994b) Air Pollution Monitoring for on-line Warning and Alarm.
Presented at the International Emergency Management and Engineering
Conference. Florida April 18-21, 1994 Lillestrøm (NILU F 7/94).

Sivertsen, B. and Bekkestad, T. (1994) Air Pollution Impact in the Border Areas
of Norway and Russia. Trends and Episodes. Presented at the 2"'' Symposium
on "Effects of air pollutants on terrestrial ecosystems in the border areas", 3-5
October 1994, Svanvik, Norway. Kjeller (NILU F 23/94)

Sivertsen, B. and Haagenrud, S.E. (1994) EU833 ENSIS '94. An environmental

surveillance system for the 1994 Winter Olympic Games. Presented at: Vision
Eureka ITEM Conference, Lillehammer, 14-15 June 1994. Kjeller (NILU
F 10/94).

Sluyter, R.J.C.F., ed. (1995) Air quality in major European cities. Part I: Scientific
background document to Europe's environment. Bilthoven/Kjeller,
RIVM/NILU (RIVM report; no. 722401004)

Tørseth, K. and Joranger, E. (1994) Overvåking av langtransportert forurenset luft

og nedbør. Atmosfærisk tilførsel 1992). Lillestrøm (NILU OR 14/94).

WHO (1987) Air Quality Guidelines for Europe. Copenhagen, World Health
Organization, Regional Office for Europe (WHO Regional Publications,
European Series No. 23).

NILU TR 11/97
~ Norwegian Institute for Air Research (NILU)
NILU P.O. Box 100, N-2007 Kjeller - Norway

REPORT SERIES REPORT NO. TR 11/97 ISBN-82-425-0913-1

TEKNISK RAPPORT ISSN 0807-7185
DATE SIGN. <f11 ~(~ NO. OF PAGES PRICE
SEPTEMBER 1997 139 NOK 180,-
TITLE
' PROJECT LEADER
Air Quality Monitoring Systems and Application Bjarne Sivertsen
NILU PROJECT NO.
Q-303
AUTHOR(S) CLASSIFICATION *
B. Sivertsen (Editor) A
CONTRACT REF.

REPORT PREPARED FOR:

Norwegian Institute for Air Research (NILU).

ABSTRACT
The elements of a modern air quality monitoring system has been presented and discussed. The monitoring
programme included sensors and instruments, data transfer, quality assurance, modelling, data presentation and
data application are all part of the system.

NORWEGIAN TITLE Integrert måle-, informasjons- og presentasjonssystem for luftkvalitet

KEYWORDS
Air quality Monitoring system Data applications
ABSTRACT (in Norwegian)

Et moderne system for luftkvalitet inneholder sensorer, instrumenter, dataloggere og dataoverføringssystem,

inkludert kvalitetskontroll, modellberegningsverktøy og grafiske presentasjonspakker. Det integrerte systemet,
samt eksempler på bruk av data er presentert i rapporten.

* Classification A Unclassified (can be ordered from NJLU)

B Restricted distribution
C Classified (not to be distributed)