COVER SHEET

Holmes, NS and Morawska, L (2006) A Review of Dispersion Modelling and its application to the
dispersion of particles: An overview of different dispersion models available. Atmospheric
Enivronment 40(30):pp. 5902-5928.

Accessed from http://eprints.qut.edu.au

Copyright 2006 Elsevier

A Review of Dispersion Modelling and its application to the dispersion of

particles: An overview of different dispersion models available.

Holmes N.S. and Morawska L.*

International Laboratory for Air Quality and Health, Queensland University of
Technology, GPO Box 2434, Brisbane QLD, 4001, Australia.

* corresponding author
International Laboratory for Air Quality and Health,
Queensland University of Technology
2 George Street
Brisbane,
QLD 4001,
Australia.
Phone: +61 7 3864 2616
Fax: +61 7 3864 1521
Email: [email protected]

Abstract
This paper provides the first review of the application of atmospheric models for
particle dispersion. The different types of dispersion models available, from simple
box type models to complex fluid dynamics models are outlined and the suitability of
the different approaches to dispersion modelling within different environments, in
regards to scale, complexity of the environment and concentration parameters is
assessed. Finally, several major commercial and non-commercial particle dispersion
packages are reviewed, detailing which processes are included and advantages and
limitations of their use to modelling particle dispersion. The models reviewed
included: Box models (AURORA, CPB and PBM), Gaussian models (CALINE4,
HIWAY2, CAR-FMI, OSPM, CALPUFF, AEROPOL, AERMOD, UK-ADMS and
SCREEN3), Lagrangian/Eulerian Models (GRAL, TAPM, ARIA Regional), CFD
models (ARIA Local, MISKAM, MICRO-CALGRID) and models which include
aerosol dynamics (GATOR, MONO32, UHMA, CIT, AERO, RPM, AEROFOR2,
URM-1ATM, MADRID, CALGRID and UNI-AERO).

Table of Acronyms used

Models
AERMOD

American Meteorological Society/Environmental Protection

Agency Regulatory Model Improvement Committee Dispersion
Model)

AEROFOR2

Model for Aerosol formation and Dynamics

AURORA

Air Quality Modelling in Urban Regions using an Optimal

Resolution Approach

CACM

Caltech Atmospheric Chemistry Mechanism

CALGRID

California Photochemical grid Model

CALINE4

California Line Source Dispersion Model

CALPUFF

California Puff Model

CAQM

Community multiscale air quality model.

CAR-FMI

Contaminants in the Air from a Road Finnish Meteorological

Institute

CBM-IV

Chemical Bond Mechanism Version IV

CIT

California/Carnegie-Mellon Institute of Technology

CMEM

Comprehensive Modal Emission Model

CPB

Canyon Plume Box

EQSAM

Equilibrium Simplified Aerosol Model

GATOR

Gas Aerosol Transport Radiation Model

GRAL

Graz Lagrangian Model

ISORROPIA

Thermodynamics Equilibrium Model (from the greek word)

MADRID

Model of Aerosol Dynamics, Reaction, Ionization and

Dissolution

MEASURE

Mobile Emissions Assessment System for Urban and Regional

Evaluation

MICRO-CALGRID Microscale California Photochemical grid Model

MISKAM

Microscale flow and dispersion model

MONO32

Multimono

OSPM

Operational Street Pollution Model

PBM

Photochemical Box Model

RACM

Regional Atmospheric Chemistry

RADM

Regional Acid Deposition Mechanism

RPM

Regional Particulate Model

SAPRC

Statewide Air Pollution Research Center

SCREEN3

Screening version of ISC3 model

SEQUILIB

Sectional Equilibrium Model

STAR-CD

Simulation of turbulent flow in arbitrary regions Computational

Dynamics

TAPM

The Air Pollution Model

TREFIC

Traffic Emission Factor Improved Calculation

UAM IV

Urban Airshed Model with Aerosols Version 4

UAM-AERO

Urban Airshed Model with Aerosols

UHMA

University of Helsinki Multicomponent Aerosol Model

UK-ADMS

UK Atmospheric Dispersion Modeling System

UNI-AERO

EMEP Aerosol Dynamics Model

URM-1ATM

Urban-Regional Model

Other terms
CBL

Convective boundary layer

CERC

Cambridge Environmental Research Consultants

CFD

Computational Fluid Dynamic

CMU

Carnegie-Mellon University

EMEP

Co-operative Programme for Monitoring and Evaluation of the

Long-range Transmission of Air Pollutants in Europe

FA2

Fraction of modelled values within a factor of 2 of measured

values

GRS

Generic reaction set

NMSE

Normalised mean square error

PM10, PM2.5, PMx

Particulate matter in the atmosphere with an aerodynamic

diameter less than 10, 2.5 m or some value x.

SOA

Secondary Organic Aerosol

UK

United Kingdom

uv

ultraviolet

ZSR

Zdanovskii Stokes Robinson

1. Introduction
Dispersion modelling uses mathematical equations, describing the atmosphere,
dispersion and chemical and physical processes within the plume, to calculate
concentrations at various locations. Whilst, there have been various review papers on
atmospheric

modelling

and

their

approaches

to

dispersion

in

street

canyons(Vardoulakis, Fisher et al. 2003) and comparisons between different models

using test meteorological data(Ellis, McHugh et al. 2001; Sivacoumar and
Thanasekaran 2001; Hall, Spanton et al. 2002; Caputo, Gimenez et al. 2003), these
have focussed on modelling gaseous dispersion.
Unfortunately, only a few studies have simultaneously measured particle
concentration with gases and the differences between the studies may be partially
responsible for the differences observed. In open sites several studies have shown
varying correlations between the concentrations of gases and particles. Monn et al.
(1997) (Monn, Fuchs et al. 1997) showed a poor correlation between the outdoor
PM10 concentrations and NO2 concentrations in an urban environment with a better
correlation between PM2.5 and NO2, although only 2 locations were studied in the
latter case. In contrast, Clairborn (1995) (Clairborn, Mitra et al. 1995) showed a good

correlation between SF6 and PM10 although only distances upto 60m from the
motorway were measured. Roorda-Knape et al. (1998) (Roorda-Knape, Janssen et al.
1998) observed that benzene, PM2.5 and PM10 showed no significant decrease in
concentration upto 300m from a major motorway. This was consistent with the small
decrease in the PM2.5 concentration observed by Hitchins et al. (2000)(Hitchins,
Morawska et al. 2000). In that study the authors observed that particle number
concentration decreased faster than NO2 concentration from a motorway. Zhu et al.
(2002)(Zhu, Hinds et al. 2002a; Zhu, Hinds et al. 2002b) showed that number
concentration of particles between

6 and 220nm correlated well with CO

concentration from a motorway. All of these studies were made in an open

environment where the wind direction was perpendicularly away from the road.
However, differences have been observed between the local dispersion of gases and
particles(Morawska

2003;

Holmes,

Morawska

et

al.

2005).

Simultaneous

measurements of CO and particle number concentration showed that CO

concentration was not significantly correlated to particle number concentration around
the site and examination of between-site comparisons with the two pollutants showed
different spatial and temporal trends. In another study of urban sites Harrison and
Jones (2005) (Harrison and Jones 2005), observed that particle concentrations
correlated only weakly with NOx, with the highest correlation observed at a curbside
monitoring location, where concentrations are less affected by dispersion. In addition,
an examination of many urban studies(Morawska 2003) has shown that the vertical
profiles of particle number concentration around buildings differed from that of gases.
These studies differ from the previous studies in that they were conducted in a more
complex environment where wind flows were heavily affected by turbulence and
emissions were not limited to a single line source. In general the studies show that in

open environments the gas and particle concentrations correlate quite well, whilst in a
more complex urban environment significant differences are observed between gas
and particle dispersion. In an urban environment where traffic emissions are the
dominant source of particles particle Van Dingenen et al. (2004)(Van Dingenen, Raes
et al. 2004) showed PM2.5 and PM10 had an R2 value of 0.95 across all sites in the
monitoring network. However, the PM10/PM2.5 ratio varied too much to propose a
single PM10/PM2.5 ratio. In the same study they observed no correlation between
annual average particle number concentration and either PM2.5 or PM10
concentrations. This is in contrast to Harrison et al. (1999) (Harrison, Jones et al.
1999) who found that in an urban measurement study hourly particle number
concentration more closely correlated with PM2.5 than PM10 measurements, although
both PM ranges showed good correlation with the hourly particle number
concentrations during the 3 month period.
Therefore, models that are designed to model the dispersion of passive scalars, such as
inert gases should be capable of modelling the PM2.5 and PM10 concentrations in
certain open environments, especially for longer averaging periods and in the larger
airshed where short term variations resulting from transient particle formation events
are evened out.
The modelling of particle number concentration involves the incorporation of aerosol
dynamics modules into dispersion models. Thus the discussion of particle dispersion
modelling must involve both a discussion of the limitations of the various dispersion
approaches to the treatment of particles and the aerosol dynamic packages used to
evaluate particle processes occurring within the plumes. To complicate the situation
further, Lohmeyer (2001) (Lohmeyer 2001)observed that concentrations calculated by
the different models differed by a factor of four and even when the same model was

employed

results

varied

between

groups.

The

agreement

with

predicted

concentrations was seen to depend on the quality of the input data.

This review will outline the different model types, looking at specific requirements for
the different spatial scales from local to regional models, and deficiencies with respect
to particle dispersion and aerosol dynamics within different scales. In addition, whilst
not being a comprehensive review of all models available a large number of models
are included in the review and the more important model parameters and inputs for
the models are listed in Tables 1a and b and Table 2.
Although several models claim to be able to model particle dispersion, without
specific treatment of particle dynamics the results are limited to calculation of particle
mass, usually in the form of PM2.5 and PM10, and cannot calculate particle number
concentration.
Furthermore particle validation studies are not available for many models. Where this
is the case the authors have attempted to highlight model performance in terms of gas
dispersion validation studies. Since several studies have shown a good correlation
between non reactive gases and particles within a larger airshed, validation studies
involving gases should be a good indicator of the performance of the model in terms
of calculating particle mass concentrations, as discussed earlier. In addition, different
averaging times between average gas and particle concentrations make comparison
difficult and mean that it is often impossible to determine whether changes between
gas and particle concentrations predicted by the model correlate so well.
A number of local and regional models exist that include extensive treatment of
aerosol dynamics. The majority of these are non commercial packages and have been
coupled to existing dispersion models in order to provide a package that is able to
model changes to particle number concentration within different size groups. This

means that the performance of these models depends on both the accuracy and
specific processes included in the dynamics module as well as the performance of the
dispersion model. It is often possible to integrate the aerosol dynamics module with
different dispersion models to adapt the coupled dispersion package to better suit the
planned study.

2. Modelling Methodology
2.1.

Box Models

Box models are based on the conservation of mass. The site is treated as a box into
which pollutants are emitted and undergo chemical and physical processes. It requires
the input of simple meteorology and emissions and the movement of pollutants in and
out of the box is allowed. The inside of the box is not defined and the air mass is
treated as if it is well mixed and concentrations uniform throughout. One advantage of
the box model is because of the simplified meteorology box models can include more
detailed chemical reaction schemes (e.g. Master Chemical Mechanism) and detailed
treatment of the aerosol dynamics, that are able to represent the chemistry and physics
of particles within the atmosphere better. However, following inputting initial
conditions a box model simulates the formation of pollutants within the box without
providing any information on the local concentrations of the pollutants. For this
reason they are unsuitable to modelling the particle concentrations within a local
environment, where concentrations and thus particle dynamics are highly influenced
by local changes to the wind field and emissions.

2.2.

Gaussian Models

Gaussian type models are widely used in atmospheric dispersion modelling, in

particular for regulatory purposes, and are often nested within Lagrangian and
Eulerian models. Gaussian models are based on a Gaussian distribution of the plume
in the vertical and horizontal directions under steady state conditions. The normal
distribution of the plume is modified at greater distances due to the effects of
turbulent reflection from the surface of the earth and at the boundary layer when the
mixing height is low. The width of the plume is determined by y and z, which are
defined either by stability classes(Pasquill 1961; Gifford Jr. 1976) or travel time from
the source. One severe limitation of plume models with regards to modelling particle
dispersion is that since the plume models use steady state approximations they do not
take into account the time required for the pollutant to travel to the receptor.
Therefore, aerosol dynamics must be calculated by post processing treatment of the
results. In addition, regional modelling generally requires the incorporation of
chemical modelling to accurately predict the formation of particles through secondary
organic aerosol (SOA) formation. Even NOx and SOx chemistry, which is
fundamental to determining particles and ozone concentrations, is often only
calculated using a simple exponential decay. More advanced models can simulate
some of the chemical transformations using post processing treatment of the
chemistry. Although most Gaussian models only consider diffusion and advection of
the pollutants more advance Gaussian models have recently been developed that
include physical processes such as deposition and fast chemical reactions.
Furthermore, the Gaussian plume equation assumes that there is no interaction
between plumes, which can become significant within urban environments.

10

Algorithms have been developed to model the chemistry and physical processes
within the plume and dispersion around buildings. The effect of wakes from buildings
can be achieved by modifying the dispersion coefficients, y and z. However, the
Gaussian equation is not able to calculate recirculation effects caused by multiple
buildings or at intersections.
Some of the restrictions implicit in the Gaussian Plume models can be overcome by
approximating the emission as a series of puffs over time, which allows the wind
speed to be varied. In this approach each puff behaves according to the Gaussian
dispersion equation and the overall contribution of the source is calculated by
integration of the individual puffs with respect to time and summation of the
contribution of individual puffs at the receptor position.
In order to calculate the concentration of pollutants over an urban area multiple source
plumes are often used. The different equations used are determined by the nature of
the source and heights of the source and receptor.
Some further limitations of the Gaussian treatment means that Gaussian models are
not designed to model the dispersion under low wind conditions or at sites close to the
source, i.e. distances less than 100m. Gaussian models have been shown to
consistently overpredict concentrations in low wind conditions(Benson 1984; Sokhi,
Fisher et al. 1998). Hybrid models, which use a combination of the Gaussian plume
and puff models, include along wind dispersion of the pollutants in order to better
estimate concentrations under low wind speed conditions(Sharan, Yadav et al. 1996;
Thomson and Manning 2001). A further limitation is a result of the simplified
treatment of turbulence and meteorology so they are best suited to calculating hourly
pollutant concentrations.

11

Since Gaussian plume equations assume a homogeneous wind field it is not

recommended that they are used for far field modelling as the meteorology is
expected to change over such large distances. Caputo et al. (2003)(Caputo, Gimenez
et al. 2003) observed that four Gaussian models calculated non zero concentrations for
the whole downwind domain and so suggested that they should be limited to distances
a few tens of kilometres from the source.

2.3.

Lagrangian Models

Lagrangian models are similar to box models in that they define a region of air as a
box containing an initial concentration of pollutants. The Lagrangian model then
follows the trajectory of the box as it moves downwind. The concentration is a
product of a source term and a probability density function as the pollutant moves
from x to x'.
Lagrangian models incorporate changes in concentration due to mean fluid velocity,
turbulence of the wind components and molecular diffusion.
Lagrangian models work well both for homogeneous and stationary conditions over
the flat terrain(Oettl, Kukkonen et al. 2001; Raza, Avila et al. 2001; Venkatesan,
Mathiyarasu et al. 2002; Tsuang 2003) and for inhomogeneous and unstable media
condition for the complex terrain(Du 2001; Hurley, Manins et al. 2003; Jung, Park et
al. 2003). It is possible to model the non-linear chemistry using either the
superimposition of a concentration grid on the domain, followed by calculation of the
concentration in each grid or the particle can be treated as an expanded box and the
photochemical module of the model applied to each box.
The meteorological data calculates the variance of the wind velocity fluctuations and
Lagrangian autocorrelation function. Since Lagrangian particle models calculate the

12

diffusion characteristics by the generation of semi random numbers they are not
confined by stability classes or sigma curves, as is the case with Gaussian dispersion
models.

2.4.

Computational Fluid Dynamic Models

Computational fluid dynamic (CFD) models provide complex analysis of fluid flow
based on conservation of mass and momentum by resolving the Navier-Stokes
equation using finite difference and finite volume methods in three dimensions.
Turbulence is classically calculated using k- closure methods to calculate the
isotropic eddy viscosity parameter present in both the momentum and pollution
transport equations, which assumes that a pollutant is diluted equally in all directions.
This treatment performs well on a flat boundary layer. However, when a stratified
boundary layer exists the closure method needs to be modified to include the Coriolis
force and reduced wind shear in the upper atmosphere, which results in an
overestimation of the eddy viscosity.
Gidhagen et al. (2004)(Gidhagen, Johansson et al. 2004) reported that different CFD
models showed good agreement in overall wind flow field but demonstrated that
although the inputs were identical the models gave large differences in velocities and
turbulence levels. Comparison with the wind tunnel data suggested that this was a
result of the closure mechanisms used by the different models.

3. Overview of models for dispersion within a street environment

A review of urban dispersion models is given by Vardoulakis et al.
(2003)(Vardoulakis, Fisher et al. 2003) so only a brief summary of the models will be
given here together with a discussion of their applicability to model particle

13

dispersion. Although there are a number of dispersion models used to calculate urban
pollutant concentrations in a local environment, some of which also include a
complex treatment of wind flow in street canyon environments, only three models
include a module to calculate particle dispersion.

3.1.

Models that exclude specific treatment of aerosol dynamics

Dispersion models that do not include a module for chemical formation (nucleation)
or aerosol dynamics (coagulation, condensation etc) are not capable of calculating
particle number concentration. Since the models are generally based on conservation
of mass they are capable of modelling the dispersion of particles, in terms of PM2.5 or
PM10. The models generally treat the particles in a similar way to gases and as
discussed earlier this can be dangerous depending on the averaging period and
location. However, since air quality regulations are currently based on particle mass
concentrations simple particle dispersion models are essential and so the performance
of the most commonly used regulatory models will be discussed as part of this review.

3.1.1.

Box models

AURORA (VITO, Belgium) is an integrated air quality model that has been used to
model the concentration of inert and reactive gases and particles in an urban
environment (Mensink, Colles et al. 2003). The model uses a steady state box model
to calculate the pollutant concentrations within a street canyon. The model assumes a
uniform concentration over the street but includes turbulent intermittency in the flow
from the upwind roof of the canyon. Inside the box both convection in the x and z
directions are considered.

14

The CPB (GEOMET) is an urban canyon box model that has been designed for urban
canyons with height to width ratios between 0.5 and 2. The model calculates the
average concentration of inert gases and NO2, using a simple algorithm for the
reaction of NO with ozone, within a street canyon for three different wind flows.

3.1.2.

Gaussian Models

Two of the most common models used to calculate the dispersion of vehicle emissions
are CALINE4 (California Department of Transportation) and HIWAY2 (US EPA).
Both models are based on a Gaussian plume model and so suffer from the inherent
limitations of the Gaussian equations to urban dispersion modelling over short
distances and within complex environments. In addition, their use is not
recommended for the modelling in low wind speeds. Despite these problems they
have been applied in a large number of studies and for regulatory purposes due to
their ease of use and since they do not require extensive computer power or time.
Both models treat traffic as an infinite line source divided into a series of elements
located perpendicular to the wind direction. Vertical dispersion parameters in
CALINE 4 take into consideration both thermal and mechanical turbulence caused by
vehicles, whilst HIWAY 2 only considers the effects of vehicles and ignores the effect
of thermal turbulence on vertical dispersion. In addition to the problems stated above,
Gaussian models (e.g. CALINE4 AND HIWAY2) lack the sophistication required for
modelling in street canyons as buildings can only be represented by changing the
surface roughness.

CAR-FMI (Finnish Meteorological Institute) is a Gaussian Plume model based on the

equations of Luhar and Patil (1989)(Luhar and Patil 1989). It is designed to calculate

15

the hourly concentrations of CO, NO, NO2, NOx and PM2.5 from vehicles.
Atmospheric stability is defined using Boundary layer scaling. As with the other
Gaussian models CAR-FMI is limited in its use in low wind conditions. The
horizontal and vertical dispersion parameters (y and z) included turbulence terms
from ambient wind speed, exhaust velocities and vehicles.
As with CALINE4, CAR-FMI models the dispersion of inert and reactive gases and
PM, using the discrete parcel method. However, unlike CALINE4 it contains
treatment of dry deposition for 3 particle size groups. Oettl et al. (2001)(Oettl,
Kukkonen et al. 2001) demonstrated that hourly NOx concentrations measured at a
major road in Finland agreed fairly well with model predictions by CAR-FMI and
GRAL. However, CAR-FMI was not able to predict the meandering wind flow under
low wind speed conditions.

OSPM is a semi empirical model that uses a Gaussian plume equation to derive the
direct contribution from the source and a box model to calculate the effect of
turbulence on the concentrations(Vignati, Berkowicz et al. 1999). Cross wind
diffusion within the plume is disregarded and the sources are treated as infinite line
sources. The plume expression for a line source is integrated along the path defined by
the street level wind.
The wind direction at the street level is assumed to be mirror reflected with respect to
the roof level wind. The wind speed at street level is calculated from the synoptic
wind speed and direction and surface roughness. The treatment and contribution of the
various turbulent processes within the street canyon vary depending on the synoptic
wind speed and direction and the reader is directed to Vignati et al. (1999)(Vignati,
Berkowicz et al. 1999) for a full description. The model allows for effects of the

16

turbulence on the concentrations at the windward and leeward sides of the canyon by
neglecting the direct component of the emissions for the windward side of the street.
The model assumes that the traffic emissions are uniformly distributed across the
canyon and empirically derives the effect of vehicle induced turbulence. The simple
treatment of turbulence means that the model is unable to model the intermittent
fluctuations of wind flow and is therefore not recommended for calculating
concentrations on timescales shorter than one hour. Additionally the model does not
take into account the cooling of the exhaust plume after emission, which could have a
significant effect on the formation of SOA particles.
The OSPM model was evaluated against measured data in an urban street
canyon(Kukkonen, Partenan et al. 2003) for NOx, NO2, O3 and CO. Predicted hourly
averaged concentrations showed fairly good agreement both at roof top and street
level. Whilst the correlation showed that it was possible to predict hourly
concentrations using modelled background concentrations and pre processed
meteorological data, no attempt was made to predict concentrations for shorter time
periods.
In a second study(Ketzel, Berkowicz et al. 2000) the agreement of OSPM with street
values was not as good and failed to accurately predict the effect of different wind
directions on hourly NO2 concentrations in two street canyons. However, it did
accurately calculate the contribution of vehicle traffic to the annual benzene
concentration at street level. This underlines the inability of the simplified treatment
of wind flow within the canyon to reflect short term changes in concentration.

3.1.3.

Lagrangian (and Eulerian) Models

17

GRAL (Institute for Internal Combustion Engines and Thermodynamics, Graz,

Austria) is a coupled Eulerain-Lagrangian Model designed to model the dispersion of
inert compounds within inhomogeneous wind fields. One limitation is that the model
can not take into account any chemical formation of particles (e.g. ammonium
nitrates, ammonium sulfates). The model calculates concentrations from 10 minutes
upto 1 hour for line and point sources as well as from tunnel portals within flat(Oettl,
Sturm et al. 2005) and complex terrain(Oettl, Sturm et al. 2003). The atmospheric
stability is calculated using boundary layer scaling. The model assumes a constant
plume rise in the vicinity of the tunnel portal as a function of the temperature
difference between the ambient air and the tunnel flow.
No validation studies currently exist for the use of GRAL with particles, although the
designers are currently involved with performing quite intensive simulations for PM10
and PM2.5 for cities. However, the model accurately simulated the concentrations of
SF6 during varying wind speeds from four tunnels surrounded by varied topography,
although the calculations for all three tunnels are highly influenced by the low
concentration values. In particular, the concentration measurements around the
Nimomiya tunnel show considerable disagreement(Oettl, Sturm et al. 2003). In a
recent study the model accurately predicted the mean hourly NOx concentration at
four locations around an urban tunnel. At the fifth location the model failed to
calculate the concentration since this location was heavily affected by surrounding
streets not included in the model. Despite the good performance of the model it is not
universally applicable without experience, since there are two empirical parameters
that are adjusted by the user due to traffic volume and specifics of the tunnel
locations, this could be the result of the poor agreement in parts for the Nimomiya
study.

18

3.1.4.

Computational Fluid Dynamic Models

ARIA Local is a CFD model that has been used to calculate real time dispersion of
gases and particles from buses and trains within urban environments(Moon, Albergel
et al. 1997; Albergel and Jasmin 1998). A variable resolution grid can be used, with
smallest grid sizes below 1 m, to include around 1 million nodes within the area under
study. Although, if topography is important it is recommended that equidistant cell
sizes are used for the horizontal grid. Three different turbulence models can be used
based on either a k- or Rij- approach with variable gas or fluid characteristics with
the atmospheric stability based on the Pasquill categories. Pollution sources include
point, line, area and volume releases with the emission generated either as a
continuous or volumic release. In addition, the fluid properties of the gases can be
adjusted to allow for either buoyant or dense gases. Although not implicitly included
in the model the effects of vehicle induced turbulence can be included by adjustment
of the model parameters. Chemical transformations can be modelled using a post
processing module.

MISKAM is a microscale dispersion for use in built up urban environments.

Typically, the domain used is around 300 x 300 m using 60 non equidistant grid cells
in each direction, although a larger domain can be modelled. Buildings are treated as
blocks and the model does not allow steep topography or include thermal effects,
buoyant releases or chemical reactions. The modelling of neutral and stable
atmospheric conditions are possible through the use of a turbulent mixing factor. A
comparison of annual mean concentrations of an arbitrary pollutant generated within a

19

wind tunnel with the model results, in a 1.2 x 1.6 x 0.14 km grid, show an excellent
agreement (R2=0.97) although the agreement decreased if the concentrations within
the city centre are included (R2=0.79) (Balczo, Farago et al. 2005). The model has the
ability to use a geometrically progressive grid to allow more refined resolution in
certain areas.

MICRO-CALGRID (Stern and Yamartino 2001)is an urban canopy scale

photochemical model that uses the flow fields and turbulence generated by the
MISKAM model. In addition to the features of MISKAM, described above, MICROCALGRID incorporates a traffic induced emissions model, MOBILEV(Fige 1997),
and the horizontal and vertical advection and diffusion schemes, a full resistance
based parameterisation of dry deposition and chemical reaction schemes, SAPRC and
CBM-IV, from CALGRID (discussed later in Section 4.4). In addition, the model
allows treatment of vehicle induced turbulence through adjustment of the total kinetic
energy of the model cells that is produced by a vehicle as it moves through the air.
Although no statistical or graphical evidence is provided, the authors state that the
behaviour of TSP was well reproduced by the model.

3.2.

Local Aerosol models involving detailed treatment of Aerosol dynamics

GATOR(Jacobson 1997) has the option of using either a moving size or stationary
size particle dynamics module coupled to a Eulerian dispersion approach to calculate
the dispersion of gases and aerosols in urban(Jacobson 1996) /and meso- scale(Lu,
Turco et al. 1997) (Jacobson 2001) environments. New particles are generated
through homogeneous nucleation that uses parameterizations for the sulphuric acid
and water reaction derived over marine environments (Pandis, Russell et al. 1994) and

20

a recent parameterisation developed for humidities greater than 60%(Fitzgerald,

Hoppel et al. 1998). Coagulation is calculated using a semi-implicit algorithm that
includes

Brownian

diffusion

(Fuchs

1964),

convective

Brownian

enhancement(Pruppacher and Klett 1997), gravitational collection(Pruppacher and

Klett 1997), turbulent inertial motion and turbulent sheer(Saffmann and Turner 1956).
Condensation and evaporation are calculated from the gas-surface equilibrium
corrected for the Kelvin effect with the liquid water content of the aerosol predicted
by the Zdanovskii, Stokes, Robinson (ZSR) method(Robinson and Stokes 1965),
including the effects of aqueous phase dissociation of inorganic and organic species.
Only dry deposition is modelled using a resistance type approach (McRae, Goodin et
al. 1982; Russell, Winner et al. 1993; Seinfeld and Pandis 1998). In addition, the
model calculates the solar irradiance that is vital for the calculation of
photodissociation from scattering and absorption curves of the particles and gases.

Pohjola et al. (2003)(Pohjola, Pirjola et al. 2003) and Gidhagen et al.

(2004)(Gidhagen, Johansson et al. 2004) recently published studies in which they
examined the particle dispersion in an urban city environment. Both studies used the
particle dynamic model MONO32 but coupled it with different dispersion models,
either a simple plume model OSPM(Vignati, Berkowicz et al. 1999) or a CFD
dispersion model STARCD(Gidhagen, Johansson et al. 2004) to calculate the number
concentration, size distribution and chemical composition of particles. In a study
within a road tunnel Gidhagen et al. (2003)(Gidhagen, Johansson et al. 2003), using
STARCD, simulated very well the particle number concentration from vehicle traffic.
The results showed excellent agreement with the measured data when velocity based
emission factors were used. The agreement decreased when constant vehicle emission

21

factors were used though the correlation was still good. A further study within a street
canyon(Gidhagen, Johansson et al. 2004) using MONO32 accurately calculated the
hourly particle concentration of particles between 7 and 450 nm using wind speed and
direction measured 10m above the roof tops and temperature and rainfall
measurements. These simulations also demonstrated that traffic induced turbulence
was important to street level dispersion. No significant difference was observed, with
and without the influence of vehicle induced turbulence, at wind speeds greater than 5
m s-1. However, at wind speeds of 2 m s-1 the total number concentration decreased by
10000 cm-3 as a result of vehicle induced turbulence.
The model was able to calculate the main features of the change of NOx concentration
on both sides of the street with wind direction. Particle concentrations showed good
agreement with the measured hourly concentration when traffic induced turbulence
and changes in aerosol concentrations due to coagulation and deposition were
included.
These studies confirm that MONO32 in conjunction with a CFD model can be used to
accurately predict aerosol dynamics of particles emitted from vehicles and shows the
improvement in the calculation when velocity based emission factors are used for
urban modelling. Recently, several new models capable predicting changes in tailpipe
emissions as a result of changes in operation, such as acceleration, or a change in
gradient have been developed including MicroFac (US EPA)(Singh, Huber et al.
2003), MEASURE (Georgia Tech), CMEM (UC Riverside) and TREFIC (ARIA
Technologies).
Using MONO32, Pohjola et al. (2003)(Pohjola, Pirjola et al. 2003) examined the
dispersion of four size sections within 25 seconds after emission. They simulated the
effects of the various processes on particle number concentration with and without

22

dilution of the plume. Neither binary nucleation nor ternary nucleation affected
particle number during the timescale of the simulation. When dilution was excluded
the particle number concentration decreased by an order of magnitude through the
effects of coagulation and increased slightly by condensation. However, when dilution
was included in the calculation, coagulation had a negligible effect on total particle
number although the number of Aitken nuclei mode particles decreased slightly and
condensation was only important when the gas phase concentration of the organic
compounds exceeded 1010 cm-3.

Korhonen et al. (2004)(Korhonen, Lehtinen et al. 2004) developed a size segregated

aerosol dynamics model, UHMA, designed to include treatment of aerosol dynamics
with a focus on new particle formation and growth. The particles are size segregated
based on the volume of the particle core, which is composed of a mixture of sulphuric
acid, water soluble organics and a variety of insoluble components. The particle
composition within each size category is identical although particle composition
varies between different size groups. The organic fraction of the particles is calculated
using a lumped description of the compounds, the properties of which can be adjusted
by the user depending on the particular conditions.
Nucleation incorporates both binary (Vehkamaki, Kulmala et al. 2002)and ternary
nucleation(Napari, Noppel et al. 2002) depending on the atmospheric ammonia
mixing ratio, with binary nucleation scheme used for ammonia concentrations lower
than 0.1 ppt.
Growth of the particles depends on both coagulation and condensation onto the
particles. The treatment of coagulation in the model is based on Brownian motion,

23

which is the major factor responsible for coagulation of submicrometre particles, and
is recalculated as a function of particle size at each time step.
Particle growth also includes condensation of low volatile organics onto the particle
following Nano-Khler theory and adsorption of ammonia and water at each time step
based on the equilibrium between the particle sulphate and water soluble organic
content and the corrected ZSR approach based on hydroscopicity measurements made
in Finland respectively.
Dry deposition of the particles is performed by applying the size dependant treatment
of deposition velocities from Rannik et al. (2003)(Rannik, Aalto et al. 2003).
The performance of the UHMA model has been validated in a number of
studies(Pirjola, Kulmala et al. 1999; Korhonen, Lehtinen et al. 2003). In addition,
Korhonen et al. (2004)(Korhonen, Lehtinen et al. 2004) investigated the model with
respect to a new particle formation event similar to particle formation events observed
over a forest(Makela, Koponen et al. 2000). They show that the model predicts well
the total particle number with the retracking the moving centre method best describing
temporal growth. However, the model over predicted the total number concentration
in the morning and failed to predict a sharp increase in particle number concentration
in the afternoon; the failure in the latter case was explained as a result of the air mass
properties at the measurement site.
In a second study Korhonen et al. (2003)(Korhonen, Lehtinen et al. 2003) examined
the ability of two modelling approaches commonly used to represent particle size
distribution within atmospheric modelling, to simulate new particle formation and
growth.
They found that the fixed sectional approach was better able to predict the total
particle concentration than the monodisperse approach, as used in MONO32, since the

24

monodisperse approach was unable to model the coagulation of newly formed

particles. The authors observed that when the number of size sections in the fixed
sectional approach is reduced the ability to accurately resolve changes in the particle
size distribution decreased. However, the greater computational power required for
the higher size resolution approaches within the sectional method means such an
approach cannot be used in large scale dispersion studies. The simpler monodisperse
method can be applied to larger scale studies as the authors demonstrated that it
adequately estimated total particle number concentration and median particle size of
the different modes

4. Overview of Urban and Regional Scale dispersion models

There are several regional dispersion models that calculate PM10 and PM2.5
concentrations without calculating the particle size distribution. Many of these are
used for regulatory purposes such as CALPUFF and TAPM(Hurley, Manins et al.
2003). Several larger scale models exist designed to model the aerosol dynamics
within an urban airshed and regional scale, including the Urban Airshed Model with
Aerosols (AERO-UAM IV), MADRID(Zhang, Pun et al. 2004), AEROFOR2(Pirjola
and Kulmala 2001), Air Quality Model with Aerosols (Pai et al. 2000)(Pai,
Vijayaraghavan et al. 2000), the California/Carnegie-Mellon Institute of Technology
models of Meng et al. (1998)(Meng, Dabdub et al. 1998) and Pilinis and Seinfeld
(1988)(Pilinis and Seinfeld 1988) and the Regional Particulate Model(Binkowski and
Shankar 1995) (include other models). These models use a separate module to
describe the aerosol dynamics coupled to a host air quality model, which is used to
model the particle dispersion.

25

The treatment of aerosol dynamics is separate from the dispersion modelling and if
the dynamics module can be integrated into a new dispersion model it is possible to
incorporate the same aerosol dynamics approach using different dispersion strategies.
This means that the comparison of air quality validation studies is difficult as the
particle concentration is not only affected by the treatment of the aerosol dynamics
but also a function of the particle dispersion.

4.1.

Models that exclude specific treatment of aerosol dynamics

4.1.1.

Box Models

The Photochemical Box Model (PBM, US EPA) is an extension of simpler box

models that simulates photochemical smog at an urban scale. Similar to other box
models it has a fixed area, typical horizontal dimensions are on the order of 10-50 km,
but unlike other box models it has a variable boundary height between 0.1 and 2 km,
consistent with the observed diurnal variation. It is suited to deal with low and
variable wind conditions in the presence of sunlight. The urban area is represented by
one or a set of cells within which the hourly concentrations of hydrocarbons and
ozone are calculated. The PBM assumes that emissions, from point, line or area
sources, are homogeneously distributed across the surface of the box and that the
volume within the box is well mixed. After inputting the initial pollutant
concentrations, hourly wind speeds, emission fluxes of CO, NOx and HCs the model
uses an extensive chemical reaction scheme and photolysis rates in junction with solar
irradiance to simulate the formation of pollutants within the box.

4.1.2.

Gaussian Models

26

AEROPOL is a steady state dispersion model for inert gases and particles upto 100
km from the source. It can also be used for local scale dispersion. It is only applicable
to flat terrain, although treatment of building effects are included. The model includes
an algorithm for plume rise based on the equations developed by Briggs
(1975)(Briggs 1975). The model calculates wet deposition as a function of
precipitation amount and dry using the deposition velocity approach. Atmospheric
stability is calculated as functions of the Pasquill stability and the authors recommend
it is used for dispersion in nearly neutral conditions and long term averages. In a
comparison with the results from the Copenhagen data set, which corresponded to an
elevated release within an urban environment in a neutral or slightly unstable
atmosphere(Kaasik and Kimmel 2003) the modelled concentrations correlated well
with the measured concentrations (R2=0.64). However, the results correlated worse
with the data set than the ADMS-UK but had a smaller tendency to calculate extreme
deviations, as represented by the fraction of the values within a factor of 2 of the
measured value (FA2) and NMSE values. The ADMS-UK comparison was performed
at a much earlier state of development than its current state and so the ADMS-UK is
currently at a more advanced development stage than AEROPOL. AEROPOL was not
specifically applied to the dispersion of particles but the authors claim that the model
is applicable to the dispersion of particles from stacks, vehicles and area sources.

CALPUFF is a multi-layer non steady state puff dispersion model designed to model
the dispersion of gases and particles using space and time varying meteorology based
on similarity equations, turbulence, emission strengths, transformation and removal. It
is able to model four different source types: point, line, volume and area using an
integrated puff formulation incorporating the effects of plume rise, partial penetration,

27

buoyant and momentum plume rise, stack effects and building effects using either the
Schulman-Scire(Schulman, Strimaitis et al. 2000) or Huber-Snyder methods. The
model calculates dry deposition using the resistance method with inputs for deposition
velocities and wet removal using a scavenging coefficient approach as a function of
precipitation intensity and type. Chemical transformations within the plume are based
on the MESOPUFF method, which is a pseudo first order chemical mechanism for
SO2, and is able to include user defined diurnal cycles of transformation rates. As a
result CALPUFF is not recommended for use in estimating the impact of NOx and
SO2 on secondary PM formation less than 10km from the source. The model does not
include any modelling of the particle dynamics. It provides hourly calculations of gas
and particle concentrations from multiple emission sources in terms of particle mass
but does not examine particle number concentration or size distribution.
CALPUFF has been used in a number of studies to investigate gas dispersion(US EPA
1998; Elbir 2003) and has been recently used to simulate a particle pollution episode
that occurred during the winter over Christchurch(Barna and Gimson 2002).
Validation studies showed good correlation with the two gas studies. Also the
predicted hourly PM10 concentrations agreed well (Index of agreement, IA ranged
from 0.67 to 0.87) with measured concentrations during a week in winter over
Christchurch. In a study of the dust blown from erosion sources within the Mexico
City basin(Villasenor, Lopez-Villegas et al. 2003) various levels of agreement were
observed

between

the

modelled

and

measured

data.

Villasenor

et

al.

(2003)(Villasenor, Lopez-Villegas et al. 2003) concluded that the days with poorer
correlations were a result of a different source. In a second study(Villasenor,
Magdaleno et al. 2003), CALPUFF failed to predict the SO2 concentrations in a
complex environment among several gas and oil exploration and production sites in

28

south east Mexico. The model underpredicted the SO2 concentrations and also showed
poor temporal agreement. In general, CALPUFF showed reasonable agreement with
pollutant concentrations in the validation studies and discrepancies appeared to the
result of unknown sources. However, due to the inherent limitation CALPUFF is not
recommended for calculation of timescales shorter than 1 hour or where dispersion is
heavily influenced by turbulence such as in an urban environment.

AERMOD (AMS/US EPA) is a near field steady state Gaussian plume model based
on planetary boundary layer turbulence structure and scaling concepts, including
treatment of both surface and elevated sources over both simple and complex terrain.
It is able to model multiple sources of different types including point, area and volume
sources. In the stable boundary layer the distribution is assumed to be Gaussian in
both the horizontal and vertical directions. However, in the convective boundary layer
(CBL) the vertical distribution is described using a bi-Gaussian probability density
function, developed by Willis and Deardorff (1981)(Willis and Deardorff 1981),
whilst the horizontal distribution is again considered to be Gaussian in nature.
AERMOD is able to model buoyant plumes and incorporates a treatment of lofting,
whereby the plume remains near the top of the boundary layer before mixing with the
CBL. In general, Gaussian models are limited to treatment of flows over a simple
terrain however, AERMOD incorporates a simple method to approximate flows over
complex terrain (Snyder et al. 1985)(Snyder, Thompson et al. 1985).
The atmosphere is described by similarity scaling relationships using only a single
measurement of surface wind speed, direction and temperature to predict vertical
profiles of wind speed and direction, temperature, turbulence and temperature

29

gradient. The model does not include dry or wet deposition of gases and only includes
a simple treatment of dry deposition using a reflection algorithm.
Whilst AERMOD is designed to model particle dispersion it has currently only been
used to investigate gas phase dispersion. One gas phase study investigated it ability to
model dispersion of an inert tracer in an urban environment. Venkatram simulated the
emission from a small source on top of a building in an urban area SF6 was released
from a line source from the top of a trailer in a car park(Venkatram 2003). He
observed that AERMOD over predicted average 30 minute concentrations at the upper
end and underpredicted concentrations at the lower end of the measured
concentrations at 24 receptor locations. However, AERMOD agrees within a factor of
two of most of the middle concentrations. The agreement between modelled and
measured concentrations at the closest receptors both in front of and behind the source
was poor, especially at nighttime. The correlation improved with distance and showed
better agreement with most of the data within a factor of two of the measured
concentrations.

UK-ADMS is a UK regulatory model developed to model the dispersion of buoyant

or neutrally buoyant particles and gases(Carruthers, Holroy et al. 1994). The model
predicts the boundary layer structure using the similarity scaling approach in a similar
method to Berkowicz et al. (1986)(Berkowicz, Olesen et al. 1986). The model uses an
advanced Gaussian approach with a normal Gaussian distribution in stable and neutral
conditions whilst the vertical dispersion is approximated by two different Gaussian
distributions in a CBL. The treatment of the reflection of the plume of the surface of
the earth is similar to other Gaussian models. ADMS calculates the plume rise based
on temperature differences between the atmosphere and the emitted plume and

30

horizontal and vertical momentum fluxes including the possibility for entrainment of
the plume and escape through the inversion at the top of the boundary layer.
The dry deposition of particles is modelled as a function of gravitational settling and
deposition velocity with respect to aerodynamic, sub-layer and surface resistances.
Wet deposition is approximated using a washout coefficient derived from the
precipitation rate.
ADMS-Urban (and some other advanced Gaussian plume models) include buildings
downwash algorithms and can model the effect of buildings, near wake recirculation
and changes in the plume centre line due to streamline deflections from the buildings.
However, the description of the canyon is limited and alignment of the canyons
restricted limiting its application to urban particle modelling. The changing wind flow
over complex terrain is calculated using FLOWSTAR, an advanced airflow model
developed by CERC. Carruthers et al. (1988)(Carruthers, Hunt et al. 1988) have
shown that FLOWSTAR models the flow well between tens of metres upto several
kilometres typically for gradients between 1 in 2 (upwind slopes and hill summits)
and 1 in 3 locally in hill wakes.
Hanna et al. (2001)(Hanna, Egan et al. 2001) compared the results of ADMS and
AERMOD to five sets of field measurements, which represent a cross-section of
scenarios common in modelling studies. In general both models performed well for all
scenarios; however, there were some significant discrepancies. Following a ground
level emission both ADMS and AERMOD underpredicted the concentration by a
factor of three close to the source due to downwash effects of nearby tanks. Overall,
both ADMS and AERMOD tended to underpredict the mean and maximum
concentrations.

31

Carruthers et al. (2000)(Carruthers, Edmunds et al. 2000) compared the results of

ADMS to measurements from urban and industrial locations in London, Ireland and
Wales. In the urban environment they modelled the emissions within a 1 km grid from
major point sources and roads with more than 25000 vehicles per day. ADMS
predicted NOx concentrations agreed well with the measured values but tended to
underpredict at nighttime and in winter. Predicted SO2 concentrations correlated
poorer and were observed to be very sensitive to wind direction. In the industrial case
again NOx concentrations correlated well with the measured values except during low
wind speed conditions. However, the PM10 and benzene concentrations were both
significantly underpredicted. The authors suggest that this could be due to emission
sources or strengths being poorly defined and the exclusion of periodic releases from
the modelling. In addition, the agreement of the predictions with the modelled data
improved when the complex terrain surrounding the site was included in the model
description.

In Belfast the model again significantly underpredicted the PM10

concentrations. Although the authors did not identify the cause of the discrepancy but
suggested that they may result from regional variations in the background
concentrations or the use of incorrect emission factors as they observed that domestic
and traffic sources contributed significantly to the PM10 concentrations.
A recent comparison(Riddle, Carruthers et al. 2004) between FLUENT (a CFD
model) and ADMS to predict dispersion from an isolated stack in neutral conditions
over flat terrain showed that the Lagrangian particle approach within FLUENT gave
similar results to ADMS but required much greater processing time. However, the
authors stressed that the CFD models were more appropriate for situations in complex
environments than ADMS.

32

SCREEN3(USEPA 1995) is a single source Gaussian plume model that is used for
regulatory purposes to calculate the concentrations upto 50km from industrial
emissions for worst-case scenarios. It is capable of modelling the dispersion of point,
area and volume sources, the latter two through a numerical integration and virtual
point source approach respectively. The atmospheric stability is calculated from
Turner stability classes (similar to the Pasquill classification) and uses 10 m wind
speeds to calculate the horizontal wind speed by correcting wind speeds at heights
above 10m using a power law. The model incorporates an algorithm to calculate the
building downwash effects for both far wake and near wake regions based on the
Schulman-Scire(Schulman and Scire 1993) and Huber-Snyder schemes. The model
uses results from Hosker (1984)(Hosker 1984) to calculate recirculation within a
cavity. The calculated concentration is a function of building area, wind speed and
source strength, and therefore sensitive to building orientation.
The effect of inversion break up is based on procedures in the Workbook of
Atmospheric Dispersion Estimates (Turner 1970) and includes considerations due to
plume rise but ignores the effects of elevated terrain. The calculation assumes a stable
wind category and a fixed wind speed of 2.5 m s-1. A similar treatment is applied to
the shoreline fumigation and the maximum ground level shoreline fumigation is
assumed to occur where the top of the stable plume intersects with the top of the well
mixed thermal boundary layer. Buoyancy plume effects are based on the treatment of
plume rise developed by Briggs (1975)(Briggs 1975) and used to adjust the vertical
and horizontal dispersion coefficients.
SCREEN3 can calculate the effect of simple elevated terrain and also the 24 hour
concentration due to plume impaction in complex terrain using the VALLEY module
in which the receptors are located above the stack release height(Burt 1977) assuming

33

a stable atmosphere and fixed wind speed. It is claimed that the model can calculate
the concentrations in flat or elevated simple terrain upto 100km, although the inherent
limitations of the Gaussian plume equation mean that any estimations must be treated
with extreme caution due to changes in wind field strengths and chemistry within the
plume over this distance. No validation studies exist for SCREEN3; however
validation studies of SCREEN(Mehdizadeh and Rifai 2004) and SCREEN2(Patel and
Kumar 1998) show poor agreement with average measurements as they are designed
to predict maximum hourly concentrations for worst case scenarios, in which the
winds are not equally distributed from all directions. SCREEN3 is an updated version
of SCREEN; however, the only algorithm that has been added that will affect the
dispersion calculation is an alternative building downwash algorithm(Schulman and
Scire 1993), which is unlikely to significantly affect the predictions of the validation
studies discussed above.

4.1.3.

Eulerian and Lagrangian Models

TAPM is an Eulerian grid based regional dispersion model that includes a Lagrangian
particle mode for near source concentrations.
The atmosphere is treated as an incompressible non-hydrostatic fluid with the
horizontal wind components determined from the momentum equations. It includes
treatment of cloud processes and boundary layer parameterisation using similarity
scaling and a k- solution to turbulence. Surface boundary conditions include changes
to surface temperature and moisture for different soil and land use types based on the
treatment by Kowalcysk et al. (1991)(Kowalcysk, Garratt et al. 1991).

34

Dry deposition is treated using a resistance method described by Physick and Garratt
(1995)(Physick and Garratt 1995) in which the scalars behave like heat in terms of
roughness and stability function with surface resistance based on surface type.
Wet deposition is only included for highly soluble gases and particles with the
partitioning calculated as a ratio of the liquid-rain water volume fraction.
Gas-phase photochemistry is based on the generic reaction set (GRS), the semiempirical mechanism developed by Azzi et al. (1992)(Azzi, Johnson et al. 1992)
including the hydrogen peroxide modification(Venkatram, Karamchandani et al.
1997) and gas and aqueous phase reactions of SO2 and particles based on Seinfeld and
Pandis (1998)(Seinfeld and Pandis 1998). However, no aerosol dynamic module is
included to describe changes to particle size distribution or particle number
concentration. Comparison of the modelled particle mass concentrations with
measured data showed good agreement for average and maximum particle
concentrations in a year long study in Melbourne(Hurley, Manins et al. 2003).
TAPM was evaluated against two of the model validation kit studies from
Indianapolis and Kincaid(Luhar and Hurley 2003), which simulated typical rural and
urban concentrations in flat terrain. When compared with CALPUFF and AERMOD,
TAPM performed as well and the agreement was even better when the observed
winds were assimilated. The authors concluded that this showed that TAPM provides
an accurate prediction of the local meteorology. From the results it was observed that
TAPM tended to predict too low concentrations in nighttime, stable or neutral
conditions and slightly too high concentrations during daytime convective or neutral
conditions. Also locations of the maxima were slightly wrong during low wind events
due to difficulties predicting the meandering of the flow.

35

A year long study(Hurley, Manins et al. 2003) of hourly averaged concentrations of

O3, NO2, PM10 and PM2.5 was performed comparing the predictions with
concentrations measured across Melbourne using a detailed emission inventory for
vehicle, commercial, domestic and biogenic sources.
They observed that TAPM tended to underpredict daily PM10 and PM2.5
concentrations averaged across all sites by about 13%. No correlation between hourly
concentrations was made due to the fact that particle air quality standards were based
on 24hr averages.
Despite slightly underpredicting annual NO2 average the model showed excellent
agreement with the maximum NO2 concentrations and very good agreement with the
O3 and particle concentrations. The underprediction of the NO2 was due to differences
in the winter and nighttime concentrations. The results also showed that TAPM
accurately predicted concentrations even when no meteorological data was taken.
In a further study (Hurley, Blockley et al. 2001), TAPM predicted hourly wind and
temperature values agreed with observed values at each site (IA = 0.84 and 0.96
respectively). SO2 concentrations showed good correlation with the measured
concentrations and a modelled annual average of 94 g m-3 compared excellently with
the observed average over all sites of 95 g m-3.

ARIA Regional model has been developed in order to analyse the dispersion of gases
and particles, coming from industrial, transportation and area sources, upto 1000km
with a resolution of between 1 km and 10 km. ARIA can process multi- and single
constituent isothermal and non isothermal gas flows as a function of the
thermodynamic properties of the gases.

36

The meteorological model incorporates a turbulence and deposition processor and is

able to calculate wind flows over simple and complex terrain from microscale to
synoptic scale that are based on the solution of the atmospheric motion equation and a
closure model for the Reynolds stresses(Pielke, Cotton et al. 1992). The mean wind
velocity is calculated using a mass consistent model to generate 3-D wind and
temperature profiles. The treatment of turbulence uses the parameterisation approach
of Hanna 1982(Hanna 1982) based on similarity scaling.
ARIA Regional model uses two different theoretical approaches allowing the user to
choose the most suitable dispersion model for the application: FARM model, which is
based on Eulerian approach and SPRAY which is based on Lagrangian approach.
FARM is used to calculate concentration and deposition of reactive emissions
including photochemistry gases and particles between 50 and 1000 km, while SPRAY
is proposed to determine concentration and deposition of non-reactive emissions over
complex terrain and focuses on particle emissions. The model calculates dispersion
using either a one or two way nesting within multiple grids.
The model incorporates treatment of the thermodynamic equilibrium between the gas
and condensed phases and includes treatment of wet and dry deposition and
radioactive decay.
SPRAY model is a Lagrangian particle model capable of calculating dispersion for
multiple sources within micro to regional scales based on the generalised Langevin
equation for inhomogeneous and non Gaussian turbulence(Thomson 1987). The
model has been developed and used to study the dispersion of passive pollutants in
complex terrain(Nanni, Riva et al. 1996; Anfossi, Desiato et al. 1998; Carvalho,
Degrazia et al. 2002; Gariazzo, Pelliccioni et al. 2004). SPRAY has been

37

updated(Ferrero, Anfossi et al. 2000) to include the Gram-Charlier probability density

function to solve the Fokker-Plank equation.
A recent study of SPRAY(Gariazzo, Pelliccioni et al. 2004) showed that despite
accurately calculating the wind speeds, although some discrepancies were observed in
the frictional velocity, agreement was in general very good. This was thought to be
result of problems of the model to take vertical remixing of the atmosphere into
account. Another problem of the model was the calculation of daytime turbulence,
which is strongly affect by thermal convection. Therefore, some differences existed in
the agreement although the model was able to reproduce the general behaviour of the
diurnal turbulence cycle. During the study period the model in general showed
reasonable agreement with measured NOx and SO2 concentration with a few major
exceptions. These occurred during periods corresponding to upwind conditions when
the modelled emissions were not expected to impact on the measurement locations.

4.2.

Regional Aerosol models involving detailed treatment of Aerosol dynamics

The CIT model, designed to model dispersion within an urban airshed, incorporates
the aerosol model of Pilinis and Seinfeld (1988)(Pilinis and Seinfeld 1988). The
model uses a sectional approach to particle size distribution with three size sections
between 0.05 and 10 m and the aerosols composed of a mixture of organic and
inorganic compounds. Nucleation was assumed to occur using only classical theory of
binary nucleation involving sulphuric acid and water. In order to reduce the
computational requirement of mass transfer of volatile species and due to
uncertainties in ambient aerosol measurement the model assumes that the aerosols are
in thermodynamic equilibrium. They use an inorganic equilibrium model that predicts
gas phases concentration of NH3, HCl and HNO3 and aerosol phase concentrations of

38

H2O, NH4+, SO42-, NO3-, Na+, Cl-, HSO4-, H2SO4, Na2SO4, NaHSO4, NaCl, NaNO3,
NH4Cl, NH4NO3, (NH4)2SO4, NH4HSO4 and (NH4)3H (SO4)2. Gas phase chemistry
was modelled using the mechanisms of Russell et al. (1988)(Russell 1988). Secondary
aerosol formation is assumed to be from three sources: aromatics, diolefins and the
cyclic ethenes, cyclopentene and cyclohexene. The dispersion model assumes that the
atmosphere exists of 5 unequally distributed layers upto 1100m with a horizontal size
of 150 km x 400km divided into 5x5 km grid squares.

The URM-1ATM model is an updated version of the CIT model and calculates the
dispersion and chemistry of the pollutants by solving the Eulerian equations for
conservation of mass using a finite element variable transport scheme coupled to the
updated SAPRC chemical mechanism(Carter 2000; Carter 2003). Aerosol dynamics
are modelled using a sectional approach with four size groups upto 10m made up of
internally mixed atmospherically relevant particles. The equilibrium based model
ISORROPIA(Nenes, Pandis et al. 1998; Nenes, Pandis et al. 1999) is used to calculate
the growth and mass transfer of particles through condensation. The module
ISORROPIA designed to calculate the gas-aerosol partitioning of inorganic
compounds is very computationally demanding. Recently Metzger et al.
(2002)(Metzger, Dentener et al. 2002) have developed a simpler model, EQSAM, that
assumes chemical equilibrium to relate the aerosol activity coefficients to relative
humidity. Metzger et al. (2002)(Metzger, Dentener et al. 2002) showed that EQSAM
was much faster and provided comparable results to ISORROPIA using a non
iterative manner. They concluded that it provides a good alternative to ISORROPIA
in global modelling applications.

39

The model incorporates a wet deposition and scavenging process developed by

Berkowitz et al. (1989)(Berkowicz, Easter et al. 1989) to simulate the formation of
clouds and kinetic interaction with the atmospheric species. The dry deposition of the
particles uses the three resistance approach as described by Wesely (1989)(Wesely
1989).

Lurmann et al. (1997)(Lurmann, Wexler et al. 1997) coupled the AERO model
dispersion model to UAM IV, which is a similar aerosol model to that employed by
Pilinis and Seinfeld (1988)(Pilinis and Seinfeld 1988). Size distribution was
represented by eight sections between 0.04 and 10 m and assumed uniform
composition of aerosols composed of inorganic and organic compounds and elemental
carbon. Nucleation and condensation were identical to the treatment by Pilinis and
Seinfeld (1988)(Pilinis and Seinfeld 1988). Only dry deposition was considered using
the method recommended by Slinn and Slinn (1980)(Slinn and Slinn 1980) to
calculate the deposition velocities. The effects of changing season and land cover on
surface resistance were calculated using the approach of Wesely (1989)(Wesely
1989).
The model was used to reproduce concentrations measured during two summer
pollution episodes in Los Angeles in 1987. The AERO-UAM IV performed poorly
with respect to the observed NO2 concentrations generally tending to underpredict the
hourly concentrations. Modelled ozone concentrations showed the desired trends but
often over and underpredicted the observed concentrations with a mean error of
35%. The model predictions for mean daily PM2.5 mass agrees relatively well with the
observed mass. Elemental carbon and crustal material was found to make up the
major component of aerosols greater than 1.2 m with nitrate ions accounting for

40

about 50% of the mass of aerosols smaller than 310nm. The authors commented that
the good agreement between the measured and modelled PM2.5 was a result of the
overestimation of the crustal emissions since several of the individual contributions
were underestimated. Predictions of the 24 hr and 4hr PM10 mass concentrations were
higher than observed at all locations. Further examination of the individual
components of the PM10 particles showed that the model again over estimated the
crustal component but underpredicted a number of the other components of the
aerosols. The most accurate component of PM10 was nitrate which is surprising due
the underprediction of the NO2 and nitric acid, which influence the formation of
nitrate ion.

CALGRID is a Eulerian dispersion model based on the UAM-IV model with

improvements to the horizontal advection(Yamartino, Scire et al. 1989), vertical
transport, deposition and chemical transformation(Scire and Yamartino 1989).
CALGRID uses regularly spaced horizontal grid sizes between 500 m and 20km and
vertical height from 20m to 2km to create a horizontal domain between 20-1000 km
and height upto 10km to calculate hourly concentrations of both reactive and inert
gases and particles within a complex terrain. Atmospheric stability and boundary
layer height is calculated using stability categories (Briggs 1973). Vertical diffusivity
is based on convective scaling during the day and local scaling at night. Plume rise of
buoyant sources within a stable, neutral or unstable atmosphere is calculated using the
treatment of Briggs (1975)(Briggs 1975).
Emissions are generated for each cell and each species in terms of mass per unit time.
CALGRID includes the photochemical mechanism SAPRC to predict the formation
of secondary gases. The treatment of aerosols includes both primary particles and

41

SOA formation. Particles are formed using the Chemical Bond Mechanism IV (CBM
IV) containing 86 reactions and 35 species (Gery, Whitten et al. 1989). In addition,
the chemical thermodynamic model ISORROPIA to provide detailed treatment of
equilibrium and partitioning between gas, liquid and solid phases.
SOA formation is a function of reaction rate and stoichiometry of the reactions as well
as a temperature dependent equilibrium partitioning dynamics. Ultraviolet (uv)
irradiance in each cell is a function of cloud cover and includes an algorithm for
transmissivity developed by Bais et al. (1993)(Bais, Zerefos et al. 1993) and the
effects of surface albedo to provide the total uv flux in a layer. This flux is used to
calculate the photolysis rates.
Three options are available for dry deposition depending on the complexity of the
calculation. The user can choose to ignore dry deposition, define 24 hour cycles of the
deposition velocities or calculate deposition rates as a function of space and time,
based on the equations of Wesley and Hicks (1977)(Wesely and Hicks 1977) for gases
and as a function of particle size using the equations from Slinn and Slinn (1980)
(Slinn and Slinn 1980) and Pleim et al (1984)(Pleim, Venkatram et al. 1984). No
treatment is included for wet deposition.
ONiell and Lamb (2005)(O'Neill and Lamb 2005) compared the results for
CALGRID coupled to the photochemical model SAPCR97 with measured O3
concentrations. They showed that the hourly model results correlated very well with
the measured concentrations. This study demonstrated that the formation of one of the
major secondary pollutants, critical to the accurate determination of particle formation
is accurately modelled by the photochemical mechanism included in CALGRID.
Villasenor et al. (2001) (Villasenor, Claiborn et al. 2001) modelled PM10 and PM2.5
concentrations using CALGRID in industrial and residential areas. They did not

42

include the photochemical mechanism, SAPRC. The model correlated excellently

(R2=0.94) with the measured hourly PM10 values and slightly less well but still
showed a good correlation with PM2.5 values (R2=0.63). The reduced correlation of
the PM2.5 values was probably a reflection of the small influence that traffic emissions
have at the sites. However, the absence of chemical transformations would also have
more affect on the PM2.5 correlation than the PM10 since the percentage contribution
of particle formation to the PM2.5 should be greater.

UNI-AERO is an aerosol model that incorporates the EMEP dispersion model within
equidistance 50 x 50km horizontal grid cells and 20 size varying vertical layers.
Horizontal and vertical advection are determined according to schemes designed by
Bott(Bott 1989). Atmospheric Stability is calculated using similarity theory and
vertical diffusivity calculated from local Richardson numbers as a function of the
atmospheric stability.
The model includes treatment of both primary and secondary particles, although
SOAs are not currently included in the standard version. Chemistry includes a full
photochemical mechanism(Kuhn, Builtjes et al. 1998) together with ammonium
chemistry, gas and aqueous oxidation of SO to sulphate. Partitioning of semi volatile
inorganic compounds between the

gas and aerosol phases is calculated using

ESQAM, which also calculates water associated with the aerosols based on chemical
composition using the ZSR relationship(Robinson and Stokes 1965). In this way the
aerosol water content calculated depends on the mass of soluble compounds and type
of salt mixture in the particle.
UNI-AERO calculates particle mass and number concentration in four modes as a
function of aerosol chemical composition, thus allowing the user more choice in the

43

cut off of aerosol diameter in PMx. Aerosol dynamics includes treatment of binary and
ternary nucleation(Korhonen, Kulmala et al. 1999; Berndt, Boege et al. 2000),
condensation and coagulation in addition to deposition.
Dry deposition is calculated using the resistance method as described by Wesley et al
(1989) (Wesely 1989)with the velocity in each cell moderated as a function of
fractional land use within each cell. Wet deposition of soluble components is treated
using both in-cloud and sub cloud equations, which are functions of the precipitation
rates and in-cloud and sub-cloud scavenging ratios
Tsyro et al. (2003)(Tsyro 2003) has observed that UNI-AERO (described as EMEP
Aerosol model) systematically underestimated the observed PM2.5 and PM10
concentrations by between 40 and 60 percent. They concluded that the discrepancy
could result from SOA that are not included in the model. They investigated the
influence of different factors on the model calculations. Inclusion of the
photochemistry model in contrast to the simplified treatment had the largest effect
whilst the effects of deposition and aerosol dynamics had only a minimal difference
between the model performance. This confirmed that the regional modelling of PM10
can be adequately performed without inclusion of aerosol dynamics.
A second study, Tsyro et al. (2005) (Tsyro 2005), investigated whether the difference
between the modelled and observed PM2.5 and PM10 concentrations could be a result
of the water content of the aerosols. They found that the unaccounted for PM2.5 mass
at two sites correlated well with the calculated residual aerosol water. When the water
associated with the aerosol was include in the model predictions the authors observed
that at most sites the daily the agreement of PM10 and PM2.5 concentrations with the
measured values improved, though there were some notable exceptions where the
correlation worsened. They postulated that this was because of the atmospheric

44

conditions used by the model, in particular temperature and humidity. Despite the
improved agreement they showed that significant fractions of the PM2.5 and PM10
mass were unidentified. They showed that the chemical composition of the PM10
aerosol calculated by the model correlates very well (R2 was between 0.55 and 0.69)
with the measured aerosol composition.
Furthermore, the model showed good correlation with the daily PM2.5 and PM10,
except for at the Spanish sites where the large discrepancy was due to the absence of
wind blown dust in the model.

The regional particulate matter (RPM) model includes a treatment of particle

dynamics incorporated into the RADM II dispersion model(Binkowski and Roselle
2003). The model approximates the size distribution using two discrete particle sizes,
representing nuclei and accumulation modes, approximated by a lognormal
distribution centred about 0.01 and 0.07 m respectively, composed of hydrates of
ammonium sulphate and ammonium nitrate. Nucleation is again based on the
homogeneous binary nucleation of sulphuric acid and water and includes a similar
treatment of condensation, coagulation and deposition used by Pilinis and Seinfeld
(1988)(Pilinis and Seinfeld 1988). The thermodynamic equilibrium within the aerosol
phase is calculated using the model developed by Saxena et al. (1986)(Saxena,
Hudischewskyi et al. 1986).
Gas phase chemistry within the model is described by the second generation chemical
mechanism developed by Stockwell et al. (1990)(Stockwell, Middleton et al. 1990)
which has since been updated to the RACM mechanism. This includes updated rate
constants and product yields from laboratory measurements and includes the new
condensed reaction mechanism for isoprene, -pinene, and -limonene with different

45

branching ratios for alkane decay. Additional changes were made to aldehyde and
aromatic chemistry in line with more recent kinetic studies.

More recently Meng et al. (1998)(Meng, Dabdub et al. 1998) developed a three
dimensional size resolved and chemically resolved aerosol model using the CIT
dispersion model. The aerosol distribution function is based on a sectional approach,
which allows individual description of the chemical composition of the aerosols
within each size section. The model calculates the mass of the particles based on the
sum of its individual components from the equations of Pilinis (1990)(Pilinis 1990).
Based on the approach of Wexler et al. (1994)(Wexler, Lurmann et al. 1994) the
model ignores the effect of coagulation on the aerosol growth but incorporates the
binary nucleation of sulphuric acid, contrary to the treatment of condensation in the
models of Pilinis and Seinfeld (1988)(Pilinis and Seinfeld 1988) and Lurmann et al.
(1997)(Lurmann, Wexler et al. 1997) which assumed instantaneous gas-aerosol
equilibrium. Condensation is modelled dynamically using the equation proposed by
Wexler et al. (1994)(Wexler, Lurmann et al. 1994) and the dry deposition calculated
from the equation of deposition velocity from Russell et al. (1993)(Russell, Winner et
al. 1993).
In addition to the aerosol thermodynamics of the earlier models Meng et al.
(1998)(Meng, Dabdub et al. 1998) incorporated the option of calculating the inorganic
gas-aerosol equilibrium by Kusik-Meissner(Kusik and Meissner 1978) and Pitzer
methods(Pitzer and Kim 1974) with respect to variations in both temperature and
relative humidity. As with the other models the water activity is estimated by the ZSR
method(Robinson and Stokes 1965) because it is as accurate as more complex
methods and requires significantly less computer power.

46

AEROFOR2(Pirjola and Kulmala 2001) uses a sectional modelling approach of over

200 evenly distributed size sections with logarithmic distribution within each section.
The composition of the aerosols within each section can be varied for soluble, weakly
soluble and insoluble particles and through the temporal treatment of the dynamics it
is possible to follow the particle number concentration as well as composition with
time. The nucleation includes formation through both homogeneous binary nucleation
and ternary nucleation, as discussed above. The model includes a multicomponent
approach to condensation of H2SO4, H2O and organic compounds on the existing
aerosols. Condensation of sulphuric acid and organic compounds depends on the
concentration difference between the gas and surface concentrations but the
thermodynamics of the condensation of the organic compounds is not considered
since the individual identity of the compound is not specified. Changes in solubility
and size of the particles due to condensation are calculated and growth of the particles
adjusted due to hygroscopic absorption of water.
Coagulation of the particles is based on Brownian coagulation coefficients(Fuchs
1964) and redistribution of the size classification is done simultaneously with
condensation effects. Dry deposition of the particles assumes Brownian diffusion,
interception and gravitational settling rates according to Schack et al. 1985(Schack Jr,
Pratsinis et al. 1985).
The gas phase chemistry is based on the EMEP mechanism(Simpson 1992), which
includes 140 chemical and photochemical reactions for 68 compounds and requires
the initial concentration, emission rate and deposition velocity for each compound.

47

The MADRID model(Zhang, Pun et al. 2004) was coupled to the Community
multiscale air quality model (CMAQ) dispersion model in order to simulate the
dispersion within the Los Angeles Basin. MADRID uses a multiple size sectional
approach with internally mixed particles to describe the size distribution. The model
includes explicit treatment of all processes except for coagulation. A parameterized
version based on the method of McMurray and Frielander (1979)(McMurray and
Frielander 1979) is used to simulate new particle formation and condensation onto
existing particles. The thermodynamics of the inorganic aerosol species is modelled
using ISORROPIA. Two approaches to the formation of SOA have been used one
based on the more comprehensive CACM mechanism and the other on the CBM-IV
or RADM approaches with additional treatment of aromatics and biogenic volatile
organic compounds. A mixing approach similar to that described by Jacobson is used
to describe the condensation, with the Carnegie-Mellon University (CMU)
approach(Capaldo, Pilinis et al. 2000) used to calculate mass transfer of gases to
particles

following

hybrid

approach

combining

both

equilibrium

and

dynamic(Capaldo, Pilinis et al. 2000; Pilinis, Capaldo et al. 2000) methods depending
on the particle size. Dry deposition is described using the algorithm of Venkatram and
Pleim (1999)(Venkatram and Pleim 1999) and wet deposition by the original CMAQ
module(Binkowski and Roselle 2003) modified to include the effects of dissociation
reaction by use of the effective Henrys Law constant. Also included in the model are
cloud and aqueous phase processes previously used in the CMAQ model. These have
been updated to include a comprehensive chemical mechanism to describe the
aqueous phase chemistry and subroutines to estimate the activation of aerosols and
scavenging by clouds including treatment of reformation of particles after cloud
evaporation.

48

Heterogeneous reactions involving HO2, NO2, NO3 and N2O5 on the surface of
particles are included in the model either as part of the CMU bulk aqueous phase
mechanism or individually.

5. Conclusion
This paper provides the first detailed review of dispersion modelling packages with
reference to the dispersion of particles in the atmosphere. The models reviewed
included: Box models (AURORA, CPB and PBM), Gaussian models (CALINE4,
HIWAY2, CAR-FMI, OSPM, CALPUFF, AEROPOL, AERMOD, UK-ADMS,
SCREEN3), Lagrangian/Eulerian Models (GRAL, TAPM, ARIA Regional), CFD
models (ARIA Local, MISKAM, MICRO-CALGRID) and models which included
aerosol dynamics (GATOR, MONO32, UHMA, CIT, AERO, RPM, AEROFOR2,
CRM-1ATM, UNI-AERO, CALGRID, MADRID). It outlines differences between
different model types and their limitations with respect to the scales and processes
included. This review showed that considerable differences exist between the
available model packages and due to the limitations of the models in terms of
mathematical treatment of dispersion dynamics and treatment of the aerosol
processes, considerable thought has to be given to the choice of the model for each
application. Factors which are critical to the choice of the model include: the
complexity of the environment, the dimensions of the model, the nature of the particle
source, the computing power and time that is required and the accuracy and time scale
of the calculated concentrations desired. Even with the most perfect model
fluctuations in the wind flow and emission strengths mean that the results generated
are only an approximation of the actual concentrations. Restrictions imposed due to
the lack of time and computing power, in addition to the uncertainties in the

49

modelling parameters, such as emission factors and description of the atmosphere,

mean that the relative importance of the individual factors must be assessed and the
models used to provide concentrations within an appropriate degree of error and time
period.
The applicability of the models to particle dispersion modelling depends heavily on
the nature of the concentration desired. Whilst, the modelling of particle number
concentration close to the source, for example in local and urban scales, requires in
depth modelling of aerosol dynamics Tsyro et al. (2003)(Tsyro 2003) have shown that
results for the UNI-AERO model indicate that aerosol dynamics has only a minor
influence on particle mass concentrations in a larger regional scale. In addition,
without the specific treatment of the chemistry and particle dynamics the dispersion
models are best used to predict mass concentrations since they are typically based on
the assumption of conservation of mass at each timestep. Therefore, within most
approximations gas phase dispersion models seem reasonably accurate with respect to
calculating average daily and annual particle mass concentrations in simple and
regional domains.
Whilst not proposing to be a review of every model available this paper provides a
source of information of applicability of the chosen model to the desired application.
It is unfortunately not possible to rank the models in terms of best to worst table as
comparison between the models and even a single validation data set has not been
performed and studies have shown that whilst one model might perform better than an
alternate model in one study the results may be reversed in a different scenario.
Therefore, the order depends on modelling timescale required, domain environment
and nature of the emission sources. Where possible comparison has been provided
between the performance of two or more models with regards a particular validation

50

data set and the user is left to decide which data set is more appropriate to their study.
We feel that major weaknesses in particle dispersion modelling exist a result of the
lack of studies that simultaneously measure particle number concentration and
gaseous pollutant concentrations and the lack of validation studies that compare the
performances of the various models against validation data. The latter point is
probably due to the fact that most of the aerosol dynamics models are not
commercially available.

51

Name
Developer
AURORA
VITO
CPB
GEOMET
CALINE 4
Californian
Department of
Transportation
HIWAY2
US EPA

Model
Type1
B

Scale2

Grid Size

Resolution

1x1 km

NA

Source
Types3
L

NA

GP

H:100500 m

1m

GP

1m

CAR-FMI
Finnish Met.
Institute

GP

10-100 m
but upto
10km
depending
on scaling
factor
Upto 10
km

AEROPOL
Bulgaria

GP

H:
adjustable
V: Not
defined
H: 101000m
V: 100m

ADMS
CERC

3D quasi
GP

L, R

GRAL

100m20km

GATOR

L, R,
G

Upto
Global

OSPM
National
Environmental
Research
Institute,
Denmark
STAR-CD

GP/Box

NA

CFD

<1 km

ARIA Local
ARIA
Technologies

CFD

depends
on scaling
factor

PBM

Box

CALPUFF
Californian
Department of
Transportation

Multi
layer
non
steady
state
GPuff
GP

H:<50 km
V:
variable
<2 km
<200km

E/L
Bi
Gaussian

SCREEN3

TAPM
CSIRO,
Australia
AERMOD
American

Atmospheric
Stability5
NA

Turbulence6

NA

NA

1 hr, 8 hr,
Worst
case

VIT,AMB

Non
reactive
gases

1 hr

VIT,AMB

CO, NO,
NO2, NOx,
PM2.5

1 hr, 8hr,
24 hr, 1 yr

BL

VIT, AMB

P,V

G,P

1 hr

AMB

H: no
limits
V: no
limits
H: no
limits
V: no
limits

P,A,L

G, P

10 mins to
1 yr

BL

VIT
AMB

P,L

G, P

10 min to
1 hr

BL

Depends
on scale of
area
NA

P,L,A,V

G, P

1 hr to 1
yr

BL

Local
(k-L model)
Vertical
inhomogeneous
turbulence and
inhomogeneous
3D wind fields
AMB

NOx,NO2,
O3, CO PM

1 hr

NA

VIT,
Empirical wind
turbulence

P,L,A,V

G, P

1 min

BL

VIT

P,L,A,V

G, P

Real time

NA

P,L,A

NA

VIT, Local
(k-L model)
Vertical
inhomogeneous
turbulence and
inhomogeneous
3D wind fields
NA

H: no
limits
V: no
limits

P,L,A,V

G, P

> 1 hr

BL

AMB

<50km

H: no
limits
V: no
limits

P,A,V

G, P

T
Worst case
scenario
meteorology

<1000 x
1000 km

P,A,V

G, P

BL

k-

L, R

<50km

H:0.3 -30
km
V :> 10 m
H: no
limits

1hr in
simple
>24 in
complex
terrain
1 hr, 8 hr,
1 yr

P,A,V,
(L

G, P

BL

AMB

H: Upto
100 km
V: Upto 2
km
3000 grid
cells upto
50km

H:<1 m +
V:<1m +
H:<1 m +
V:<1m +

Pollutants4
CO, NO2,
SO2, PM10
NO2 and
inert gases
CO, NO2,
TSP

Output
frequency
1 hr, 24
hr, 1 yr

1 hr,
24 hr, 1 yr

Limited AMB

52

Met. Society

Steady
State GP

V: no
limits

SPRAY
ARIA
Technologies

L, R

<1-100
km

MISKAM

CFD

<300 m

MICROCALGRID

CFD

<10 km

H: 1 m to
4 km
V: 1 m to
4 km
H: 1m (60
cells in
each
direction)
V: 1m (20
cells)
H: 1m
V: 1m

treated
as series
of V)
P, L, V

G, P

1 min+

BL

P, L, V

G, P

1 min+

BL

AMB

P, L, V

G, P

1 min+

BL

VIT, AMB

NA = Not applicable
1
Model Types: B = Box, G P = Gaussian Plume, L = Lagrangian, E = Eulerian, CFD = Computational
Fluid Dynamics, GPuff = Gaussian Puff
2
Scale: L = Local, R = Regional
3
Source Types: L = Line, P = Point, A = Area, V = Volume
4
Pollutants: G = Gases, P = Particles
5
Atmospheric Stability: P = Pasquill, BL = Boundary Layer Scaling, T = Turner
6
Tubulence: VIT = Vehicle Induced Turbulence, AMB = Turbulence of Ambient Air

Table 1a. Basic Parameters for Models not containing Aerosol Dynamics modules

53

Topography

Intersections

Plume Rise

Chemistry

Aerosol
Dynamics

Building
Wake
Effects1
X

Simple

Simple

Simple

DPM

Simple

Simple

DPM

Simple

Deposition

Complex

X
Y

X
X
Y

Complex
Simple
Simple

X
X
X

Y
X

X
Y
Y
(NO-NO2O3
chemistry)

X
Y
X

Y
Y

Complex
Complex

Name
Developer

Street
Canyon

AURORA
VITO
CPB
GEOMET
CALINE 4
Californian
Department of
Transportation
HIWAY2
US EPA
CAR-FMI
Finnish Met.
Institute
AEROPOL
Bulgaria
ADMS
CERC
GRAL
GATOR
OSPM
National
Environmental
Research
Institute,
Denmark
STAR-CD
ARIA Local
ARIA
Technologies
PBM
CALPUFF
Californian
Department of
Transportation
SCREEN3

X
X

X
S-S
H-S

X
Complex

X
X

X
X

Y
X

S-S
H-S
S-S
H-S

Simple and
Complex
Complex

Y GRS

Y
Simple SO2
decay

X
Simple
(NO-NO2
conversion
model)
Y

TAPM
CSIRO,
Australia

AERMOD
American
Met. Society
SPRAY
ARIA
Technologies
MISKAM

Evaluation
version

Simple and
Complex

Y
Simplified
Glendinning
et al. (1984)
X

Simple

Simple and
Complex

MICROCALGRID

X Not included, Y included

1
Building Wake Effects: S-S = Schulman-Scire, H-S = Huber-Snyder

Table 1b. Processes included in the dispersion models not containing an Aerosol
Dynamics package

54

Nucleation1

Coagulation

B+T

Coupled to
OSPM

B+T

AERO

Coupled to
UAM-IV

GATOR

Eulerian

MADRID

Coupled to
CAQM
Sectional
Box

SOA, B
B,T

URM

Eulerian

RPM

Incorporat
ed into
RADMII

Name
Developer
UHMA
University
of Helsinki

Dispersion
model

MONO32

AEROFOR

CIT
Californian
Institute of
Technology

Deposition2

Condensation
/ Evaporation
Y

D: Y
W: X

D: Y
W: X

D: Y
W: X
D: Y
W: X
D: Y
W: Y

Moving size or
stationary size
Multiple size
sectional
200 groupings

D: Y
W: Y
D: Y
W: Y

4 groups <10
m
0.01-0.07m

D: Y
W:X

0.5-10 m

D:Y
W:X

Particle Size
method
Hybrid/ moving
centre of
retacking
methods
0.7nm-2m
4 size modes.
Monodisperse
approach 7-450
nm
0.01-10m

Particle
composition
H2SO4,
Inorganics,
Organics
None

Inorganic,
organic and
elemental
carbon.
Internally
mixed
None

Externally or
internally mixed
varying within
each size group
Internally
mixed
Ammonium
Sulphate
Ammonium
Nitrate
Organic
Inorganics

Y = process included, X = process not included

1
Nucleation: B = Binary, T = ternary, SOA = Secondary organic aerosol formation
2
Deposition: D = Dry deposition, W = Wet deposition

Table 2. Aerosol Dynamics models

55

6. References
Albergel, A. and F. Jasmin (1998). "3-D simulation of local-scale traffic pollution."
International Journal of Vehicle Design 20(1-4): 79-87.
Anfossi, D., F. Desiato, et al. (1998). "TRANSALP 1989 experimental campaign - II.
Simulation of a tracer experiment with Lagrangian particle models."
Atmospheric Environment 32(7): 1157-1166.
Azzi, M., G. M. Johnson, et al. (1992). An introduction to the generic reaction set
photochemical smog mechanism. Proceedings of the 11th International
Conference of the Clean Air Society of Australia & New Zealand, Brisbane.
Bais, A. F., C. S. Zerefos, et al. (1993). "Spectral measurements of solar UVB
radiation and its relations to total ozone, SO2 and clouds." Journal of
Geophysical Research 98: 5199-5204.
Balczo, M., T. Farago, et al. (2005). Modelling urban pollution dispersion by using
MISKAM. Proceedings der Konferenz microCAD 2005, Miskolc University.
Barna, M. G. and N. R. Gimson (2002). "Dispersion modelling of a wintertime
particulate pollution episode in Christchurch, New Zealand." Atmospheric
Environment 36(21): 3531-3544.
Benson, P. E. (1984). CALINE 4 A Dispersion Model for Predicting Air Pollutant
Concentrations near Roadways. FHWA User Guide. U. Trinity Consultants
Inc.
Berkowicz, C. E., R. C. Easter, et al. (1989). "Theory and results from a quasi-steadystate precipitation-scavenging model." Atmospheric Environment 23: 15551571.
Berkowicz, R., J. R. Olesen, et al. (1986). The Danish Gaussian air pollution model
(OLM): Description, test and sensitivity analysis, in view of regulatory
applications. Air Pollution Modeling and Its Application. V. C. De Wispelaire,
F. A. Schiermeier and N. V. Gillani. New York, Plemum: 453-481.
Berndt, T., O. Boege, et al. (2000). "Formation of new particles in the system
H2SO4(SO3)/H2O/(NH3)-first results from a flow-tube study." Journal of
Aerosol Science 31(Suppl. 1): S554-555.
Binkowski, F. S. and S. J. Roselle (2003). "Models-3 community multiscale air
quality (CMAQ) model aerosol component - 1. Model description." Journal of
Geophysical Research-Atmospheres 108(D6): -.
Binkowski, F. S. and U. Shankar (1995). "The Regional Particulate Matter Model .1.
Model description and preliminary results." Journal of Geophysical ResearchAtmospheres 100(D12): 26191-26209.
Bott, A. (1989). "A positive definite advection scheme obtained by non-linear renormalisation of the advection fluxes." Monthly Weather Review 117: 10061015.
Briggs, G. (1973). "Internal memo as reported by F.A. Gifford Jr. in Turbulent
Diffusion Typing Schemes: A Review." Nuclear Safety 17: 68-86.
Briggs, G. A. (1975). Plume Rise Predictions. Lectures on Air Pollution and
Environmental Impact Analysis. D. A. Haugen. Boston, MA, American
Meteorology Society: 59-111.
Burt, E. W. (1977). Valley Model User's Guide. U. S. E. P. Agency, U.S.
Environmental Protection Agency, Research Triangle Park, NC.

56

Capaldo, K. P., C. Pilinis, et al. (2000). "A computationally efficient hybrid approach
for dynamic gas/aerosol transfer in air quality models." Atmospheric
Environment 34(21): 3617-3627.
Caputo, M., M. Gimenez, et al. (2003). "Intercomparison of atmospheric dispersion
models." Atmospheric Environment 37(18): 2435-2449.
Carruthers, D. J., H. A. Edmunds, et al. (2000). "Use and validation of ADMS-Urban
in contrasting urban and industrial locations." International Journal of
Environment and Pollution 14(1-6): 364-374.
Carruthers, D. J., D. R. J. Holroy, et al. (1994). "Uk-Adms - a New Approach to
Modeling Dispersion in the Earths Atmospheric Boundary-Layer." Journal of
Wind Engineering and Industrial Aerodynamics 52(1-3): 139-153.
Carruthers, D. J., J. C. R. Hunt, et al. (1988). Computational model of airflow over
hills. FLOWSTAR I. Proc. Of Envirosoft., Springer Verlag.
Carter, W. P. L. (2000). Implementation of the SAPRC-99 Chemical Mechanism into
the models-3 framework, ftp://ftp.cert.ucr.edu/pub/carter/pubs/s99mod3.pdf.
Carter, W. P. L. (2003). The SAPRC-99 Chemical Mechanism and updated VOC
Reactivity Scales. http://helium.ucr.edu/~carter/reactdat.htm.
Carvalho, J. D., G. A. Degrazia, et al. (2002). "Lagrangian stochastic dispersion
modelling for the simulation of the release of contaminants from tall and low
sources." Meteorologische Zeitschrift 11(2): 89-97.
Clairborn, C., A. Mitra, et al. (1995). "Evaluation of Pm10 Emission Rates from
Paved and Unpaved Roads Using Tracer Techniques." Atmospheric
Environment 29(10): 1075-1089.
Du, S. M. (2001). "A heuristic Lagrangian stochastic particle model of relative
diffusion: model formulation and preliminary results." Atmospheric
Environment 35(9): 1597-1607.
Elbir, T. (2003). "Comparison of model predictions with the data of an urban air
quality monitoring network in Izmir, Turkey." Atmospheric Environment
37(15): 2149-2157.
Ellis, K., C. McHugh, et al. (2001). "Comparison of ADMS-Roads, Caline4 and UK
DMRB Model Predictions for Roads." 7th International Conference on
Harmonisation within Atmospheric Dispersion Modelling for Regulatory
Purposes.
Ferrero, E., D. Anfossi, et al. (2000). "Intercomparison of Lagrangian stochastic
models based on two different PDFs." International Journal of Environment
and Pollution 14(1-6): 225-234.
Fige (1997). Mobilev-Dokumentation und Benutzerhandbuch. Foschungsvorhaben
105 06 044 des Umweltbundesamts "Erarbeitun von Grundlagen fuer die
Umsetzung von 40.2 des BImSchG". Umweltbundesamt. Berlin.
Fitzgerald, J. W., W. A. Hoppel, et al. (1998). "A One-Dimensional Sectional Model
to Simulate Multicomponent Aerosol Dynamics in the Marine Boundary
Layer. 1 Model Description." Journal of Geophysical Research-Atmospheres
103: 16085-16102.
Fuchs, N. A. (1964). The mechanics of aerosols. London, Pergamon Press.
Gariazzo, C., A. Pelliccioni, et al. (2004). "Evaluation of a Lagrangian particle model
(SPRAY) to assess environmental impact of an industrial facility in complex
terrain." Water Air and Soil Pollution 155(1-4): 137-158.
Gery, M. W., G. Z. Whitten, et al. (1989). "A photochemical kinetics mechanism for
urban and regional scale computer modeling." Journal of Geophysical
Research 94: 12925-12956.

57

Gidhagen, L., C. Johansson, et al. (2004). "Simulation of NOx and ultrafine particles
in a street canyon in Stockholm, Sweden." Atmospheric Environment 38(14):
2029-2044.
Gidhagen, L., C. Johansson, et al. (2003). "Model simulation of ultrafine particles
inside a road tunnel." Atmospheric Environment 37: 2023-2036.
Gifford Jr., F. A. (1976). "Consequences of Effluent Releases." Nuclear Safety 17(1):
68-86.
Hall, D. J., A. M. Spanton, et al. (2002). "Evaluation of new generation atmospheric
dispersion models." International Journal of Environment and Pollution 18(1):
22-32.
Hanna, S. R. (1982). Applications in Air Pollution Modeling. Atmospheric
Turbulence and Air Pollution Modelling. F. T. M. Nieuwstadt and H. Van
Dop. Dordrecht, Riedel.
Hanna, S. R., B. A. Egan, et al. (2001). "Evaluation of the ADMS, AERMOD, and
ISC3 dispersion models with the OPTEX, Duke Forest, Kincaid, Indianapolis
and Lovett field datasets." International Journal of Environment and Pollution
16(1-6): 301-314.
Harrison, R. M., M. Jones, et al. (1999). "Measurements of the Physical Properties of
Particles in the Urban Atmosphere." Atmospheric Environment 33: 309-321.
Harrison, R. N. and A. M. Jones (2005). "Multisite study of particle number
concentrations in urban air." Environmental Science & Technology 39(16):
6063-6070.
Hitchins, J., L. Morawska, et al. (2000). "Concentrations of submicrometre particles
from vehicle emissions near a major road." Atmospheric Environment 34: 5164.
Holmes, N. S., L. Morawska, et al. (2005). "Spatial distribution of submicrometre
particles and CO in an urban microscale environment." Atmospheric
Environment 39(22): 3977-3988.
Hosker, R. P. (1984). Flow and Diffusion Near Obstacles. Atmospheric Science and
Power Production. D. Randerson. Washington, D.C., U.S. Department of
Energy.
Hurley, P., P. Manins, et al. (2003). "Year-long, high-resolution, urban airshed
modelling: verification of TAPM predictions of smog and particles in
Melbourne, Australia." Atmospheric Environment 37(14): 1899-1910.
Hurley, P. J., A. Blockley, et al. (2001). "Verification of a prognostic meteorological
and air pollution model for year-long predictions in the Kwinana industrial
region of Western Australia." Atmospheric Environment 35(10): 1871-1880.
Jacobson, M. Z. (1996). "Application of a sparse-matrix, vectorized gear-type code in
a new air pollution modeling system." Zeitschrift Fur Angewandte
Mathematik Und Mechanik 76: 333-336.
Jacobson, M. Z. (1997). "Development and application of a new air pollution
modeling system .2. Aerosol module structure and design (vol 31, pg 131,
1997)." Atmospheric Environment 31(7): 1097-1097.
Jacobson, M. Z. (2001). "GATOR-GCMM: A global- through urban-scale air
pollution and weather forecast model 1. Model design and treatment of
subgrid soil, vegetation, roads, rooftops, water, sea ice, and snow." Journal of
Geophysical Research-Atmospheres 106(D6): 5385-5401.
Jung, Y. R., W. G. Park, et al. (2003). "Pollution dispersion analysis using the puff
model with numerical flow field data." Mechanics Research Communications
30(4): 277-286.

58

Kaasik, M. and V. Kimmel (2003). "Validation of the improved AEROPOL model

against the Copenhagen data set." International Journal of Environment and
Pollution 20(1-6): 114-120.
Ketzel, M., R. Berkowicz, et al. (2000). "Comparison of numerical street dispersion
models with results from wind tunnel and field measurements." Environmental
Monitoring and Assessment 65(1-2): 363-370.
Korhonen, H., K. E. J. Lehtinen, et al. (2004). "Multicomponent aerosol dynamics
model UHMA: model development and validation." Atmospheric Chemistry
and Physics 4: 757-771.
Korhonen, H., K. E. J. Lehtinen, et al. (2003). "Simulation of atmospheric nucleation
mode: A comparison of nucleation models and size distribution
representations." Journal of Geophysical Research-Atmospheres 108(D15): -.
Korhonen, P., M. Kulmala, et al. (1999). "Ternary nucleation of H2SO4, NH3, and
H2O in the atmosphere." Journal of Geophysical Research-Atmospheres
104(D21): 26349-26353.
Kowalcysk, E. A., J. R. Garratt, et al. (1991). A soil canopy scheme for use in a
numerical model of the atmosphere - 1D stand alone model. C. D. o. A.
Research, CSIRO.
Kuhn, M., P. J. H. Builtjes, et al. (1998). "Intercomparison of the gas-phase chemistry
in several chemistry and transport models." Atmospheric Environment 32(4):
693-709.
Kukkonen, J., L. Partenan, et al. (2003). "Evaluation of the OSPM model combined
with an urban background model against the data measured in 1997 in
Runeberg Street, Helsinki." Atmospheric Environment 37(8): 1101-1112.
Kusik, C. L. and H. P. Meissner (1978). "Electrolyte activity coefficients in inorganic
processing." AIChE Symp. Series 173(14-20).
Lohmeyer, A. (2001). Comparison of the procedures of different modellers for air
pollutant concentrations prediction in a street canyon-The Podbielski Street
exercise, http://www.lohmeyer.de/podbi/.
Lu, R., R. P. Turco, et al. (1997). "An integrated air pollution modeling system for
urban and regional scales .1. Structure and performance." Journal of
Geophysical Research-Atmospheres 102(D5): 6063-6079.
Luhar, A. K. and P. J. Hurley (2003). "Evaluation of TAPM, a prognostic
meteorological and air pollution model, using urban and rural point-source
data." Atmospheric Enivironment 37: 2795-2810.
Luhar, A. K. and R. Patil (1989). "A General Finite Line Source Model for Vehicular
Pollution Dispersion." Atmospheric Environment 23: 555-562.
Lurmann, F. W., A. S. Wexler, et al. (1997). "Modelling urban and regional aerosols
.2. Application to California's South Coast Air Basin." Atmospheric
Environment 31(17): 2695-2715.
Makela, J. M., I. K. Koponen, et al. (2000). "One-year data of submicron size modes
of tropospheric background aerosol in Southern Finland." Journal of Aerosol
Science 31(5): 595-611.
McMurray, P. H. and S. K. Frielander (1979). "New particle formation in the presence
of an aerosol." Atmospheric Environment 13: 1635-1651.
McRae, G. J., W. R. Goodin, et al. (1982). "Development of a second-generation
mathematical model for urban air pollution." Atmospheric Environment 16:
679-696.

59

Mehdizadeh, F. and H. S. Rifai (2004). "Modeling point source plumes at high

altitudes using a modified Gaussian model." Atmospheric Environment 38(6):
821-831.
Meng, Z. Y., D. Dabdub, et al. (1998). "Size-resolved and chemically resolved model
of atmospheric aerosol dynamics." Journal of Geophysical ResearchAtmospheres 103(D3): 3419-3435.
Mensink, C., A. Colles, et al. (2003). "Integrated air quality modelling for the
assessment of air quality in streets against the council directives."
Atmospheric Environment 37(37): 5177-5184.
Metzger, S., F. Dentener, et al. (2002). "Gas/aerosol partitioning: 1. A
computationally efficient model." Journal of Geophysical ResearchAtmospheres 107(D16): 10102.
Monn, C., A. Fuchs, et al. (1997). "Particulate Matter less than 10 m (PM10) and
Fine Particles less than 2.5 m (PM2.5): Relationship between Indoor and
Outdoor and Personal Concentrations." The Science of the Total Environment
208: 15-21.
Moon, D., A. Albergel, et al. (1997). "The use of the MERCURE CFD code to deal
with an air pollution problem due to building wake effects." Journal of Wind
Engineering and Industrial Aerodynamics 67-8: 781-791.
Morawska, L. (2003). Motor vehicle emissions as source of indoor particles. Indoor
Environment. L. Morawska and T. Salthammer. Weinheim, Wiley-VCH.
XVII: 297-319.
Nanni, A., G. M. Riva, et al. (1996). "Particle model simulation of pollutants
dispersion from a line source in complex terrain." Science of the Total
Environment 190: 301-309.
Napari, I., M. Noppel, et al. (2002). "An improved model for ternary nucleation of
sulfuric acid-ammonia-water." Journal of Chemical Physics 116(10): 42214227.
Nenes, A., S. N. Pandis, et al. (1998). "ISORROPIA: A new thermodynamic
equilibrium model for multiphase multicomponent inorganic aerosols."
Aquatic Geochemistry 4(1): 123-152.
Nenes, A., S. N. Pandis, et al. (1999). "Continued development and testing of a new
thermodynamic aerosol module for urban and regional air quality models."
Atmospheric Environment 33(10): 1553-1560.
Oettl, D., J. Kukkonen, et al. (2001). "Evaluation of a Gaussian and a Lagrangian
model against a roadside data set, with emphasis on low wind speed
conditions." Atmospheric Environment 35(12): 2123-2132.
Oettl, D., P. J. Sturm, et al. (2005). "Evaluation of GRAL for the pollutant dispersion
from a city street tunnel portal at depressed level." Environmental Modelling
& Software 20: 499-504.
Oettl, D., P. J. Sturm, et al. (2003). "Dispersion from road tunnel portals:comparison
of two different modelling approaches." Atmospheric Environment 37(37):
5165-5175.
O'Neill, S. M. and B. K. Lamb (2005). "Intercomparison of the community multiscale
air quality model and CALGRID using process analysis." Environmental
Science & Technology 39(15): 5742-5753.
Pai, P., K. Vijayaraghavan, et al. (2000). "Particulate matter modeling in the Los
Angeles basin using SAQM-AERO." Journal of the Air & Waste Management
Association 50(1): 32-42.

60

Pandis, S., L. M. Russell, et al. (1994). "The Relationship Between DMS Flux and
CCN Concentration in Remote Marine Regions." Journal of Geophysical
Research-Atmospheres 99: 16945-16957.
Pasquill, F. (1961). "The Estimation of the Dispersion of Windborne Material."
Meteorology Magazine 90(1063): 33-40.
Patel, V. C. and A. Kumar (1998). "Evaluation of three air dispersion models:
ISCST2, ISCLT2, and SCREEN2 for mercury emissions in an urban area."
Environmental Monitoring and Assessment 53(2): 259-277.
Physick, W. L. and J. R. Garratt (1995). "Incorporation of a High-Roughness Lower
Boundary into a Mesoscale Model for Studies of Dry Deposition over
Complex Terrain." Boundary-Layer Meteorology 74(1-2): 55-71.
Pielke, R. A., W. R. Cotton, et al. (1992). "A Comprehensive Meteorological
Modeling System - Rams." Meteorology and Atmospheric Physics 49(1-4):
69-91.
Pilinis, C. (1990). "Derivation and numerical solution of the species mass distribution
equations for multicomponent particulate systems." Atmospheric Environment
Part a-General Topics 24: 1923-1928.
Pilinis, C., K. P. Capaldo, et al. (2000). "MADM - A new multicomponent aerosol
dynamics model." Aerosol Science and Technology 32(5): 482-502.
Pilinis, C. and J. H. Seinfeld (1988). "Development and Evaluation of an Eulerian
Photochemical Gas-Aerosol Model." Atmospheric Environment 22(9`): 19852001.
Pirjola, L. and M. Kulmala (2001). "Development of particle size and composition
distributions with a novel aerosol dynamics model." Tellus Series B-Chemical
and Physical Meteorology 53(4): 491-509.
Pirjola, L., M. Kulmala, et al. (1999). "Formation of sulphuric acid aerosols and cloud
condensation nuclei: An expression for significant nucleation and model
comparison." Journal of Aerosol Science 30(8): 1079-1094.
Pitzer, K. S. and J. J. Kim (1974). "Thermodynamics of electrolytes - IV.Activity and
osmotic coefficients for mixed electrolytes." Journal of American Chemical
Society 96: 5701-5707.
Pleim, J., A. Venkatram, et al. (1984). The Dry Deposition Model. Volume 4
ADOM/TADAP Model Development Program. Rexdale, Ontario, Canada,
Ontario Ministry of the Environment.
Pohjola, M., L. Pirjola, et al. (2003). "Modelling of the influence of aerosol processes
for the dispersion of vehicular exhaust plumes in street environment."
Atmospheric Environment 37(3): 339-351.
Pruppacher, H. R. and J. D. Klett (1997). Microphysics of Clouds and Precipitation,
Springer.
Rannik, U., P. Aalto, et al. (2003). "Interpretation of aerosol particle fluxes over a
pine forest: Dry deposition and random errors." Journal of Geophysical
Research-Atmospheres 108(D17): -.
Raza, S. S., R. Avila, et al. (2001). "A 3-D Lagrangian stochastic model for the mesoscale atmospheric dispersion applications." Nuclear Engineering and Design
208(1): 15-28.
Riddle, A., D. Carruthers, et al. (2004). "Comparisons between FLUENT and ADMS
for atmospheric dispersion modelling." Atmospheric Environment 38(7):
1029-1038.
Robinson, R. A. and R. J. Stokes (1965). Electrolyte Solutions. London, Butterworths.

61

Roorda-Knape, M. C., N. A. H. Janssen, et al. (1998). "Air Pollution from Traffic in

City Districts near Major Motorways." Atmospheric Environment 32: 19211930.
Russell, A. G. (1988). Environmental Science & Technology 22: 1336-.
Russell, A. G., R. A. Winner, et al. (1993). "Mathematical modeling and control of
the dry deposition flux of nitrogen-containing air pollutants." Environmental
Science & Technology 27(2772-2782).
Saffmann, P. G. and J. S. Turner (1956). "On the collision of drops in turbulent
clouds." Journal of Fluid Mechanics 1: 16-30.
Saxena, P., A. B. Hudischewskyi, et al. (1986). "A comparative study of equilibrium
approaches ti the chemical characterization of secondary aerosols."
Atmospheric Environment 20(1471-1483).
Schack Jr, C. J., S. E. Pratsinis, et al. (1985). "A general correlation for deposition of
suspended particles from turbulent gases to completely rough surfaces."
Atmospheric Environment 19: 953-960.
Schulman, L. L. and J. S. Scire (1993). "Building Downwash Screening Modeling for
the Downwind Recirculation Cavity." Journal of the Air & Waste
Management Association 43(8): 1122-1127.
Schulman, L. L., D. G. Strimaitis, et al. (2000). "Development and evaluation of the
PRIME plume rise and building downwash model." Journal of Air and Waste
Management Association 50: 378-390.
Scire, J. S. and R. J. Yamartino (1989). CALGRID; A Mesoscale Photochemical Grid
Model Volume II: User's Guide, University of Iowa.
Seinfeld, J. H. and S. Pandis (1998). Atmospheric Chemistry and Physics from Air
Pollution to Climate Change. New York, Wiley.
Sharan, M., A. K. Yadav, et al. (1996). "Plume dispersion simulation in low-wind
conditions using coupled plume segment and Gaussian puff approaches."
Journal of Applied Meteorology 35(10): 1625-1631.
Simpson, D. (1992). "Long-period modelling of photochemical oxidants in Europe.
Model calculation for July 1985." Atmospheric Environment 26A: 1609-1634.
Singh, R. B., A. H. Huber, et al. (2003). "Development of a Microscale Emission
Factor Model for Particulate Matter for Predicting Real-Time Motor Vehicle
Emissions." Journal of Air and Waste Management Association 53(10): 12041217.
Sivacoumar, R. and K. Thanasekaran (2001). "Comparison and performance
evaluation of models used for vehicular pollution prediction." Journal of
Environmental Engineering-Asce 127(6): 524-530.
Slinn, S. A. and W. G. N. Slinn (1980). "Predictions for particle deposition on natural
waters." Atmospheric Environment 24: 1013-1016.
Snyder, W. H., R. S. Thompson, et al. (1985). "The structure of the strongly stratified
flow over hills: Dividing streamline concept." Journal of Fluid Mechanics
152(249-288).
Sokhi, R., B. Fisher, et al. (1998). Modelling of Air Quality around Roads. Proc of 5th
Int. Conf. On Harmonisation with Atmospheric
Dispersion Modelling for Regulatory Purposes, Greece.
Stern, R. and R. J. Yamartino (2001). "Development and first evaluation of microcalgrid: a 3-D, urban-canopy-scale photochemical model." Atmospheric
Environment 35: S149-S165.

62

Stockwell, W. R., D. Middleton, et al. (1990). "The second-generation Regional Acid

Deposition Model chemical mechanism for regional air quality modeling."
Journal of Geophysical Research-Atmospheres 95: 16343-16367.
Thomson, D. J. (1987). "Criteria for selection of stochastic models of particle
trajectories in turbulent flows." Journal of Fluid Mechanics 180: 529-556.
Thomson, D. J. and A. J. Manning (2001). "Along-wind dispersion in light wind
conditions." Boundary-Layer Meteorology 98(2): 341-358.
Tsuang, B. J. (2003). "Quantification on the source/receptor relationship of primary
pollutants and secondary aerosols by a Gaussian plume trajectory model: Part I
theory." Atmospheric Environment 37(28): 3981-3991.
Tsyro, S. G. (2003). Model performance for particulate matter. In: EMEP Status
report 1/2003. Transboundary acidification, eutrophication and ground ozone
level, Part II: Unified EMEP model performance. EMEP/MSC-W & MSC-E
Status report 1/2003 Part II. N. M. Institute. Oslo.
Tsyro, S. G. (2005). "To what extent can aerosol water explain the discrepancy
between model calculated and gravimetric PM10 and PM2.5?" Atmospheric
Chemistry and Physics 5: 515-532.
Turner, D. B. (1970). Workbook of Atmospheric Dispersion Estimates.
US EPA (1998). A Comparison of Calpuff Modeling Results to Two Tracer Field
Experiments, http://www.epa.gov/scram001/7thconf/calpuff/tracer.pdf.
USEPA (1995). Screen3 Model User's Guide. U. EPA. Research Triangle Park, North
Carolina, http://www.epa.gov/scram001/userg/screen/screen3d.pdf.
Van Dingenen, R., F. Raes, et al. (2004). "A European aerosol phenomenology-1:
physical characteristics of particulate matter at kerbside, urban, rural and
background sites in Europe." Atmospheric Environment 38(16): 2561-2577.
Vardoulakis, S., B. E. A. Fisher, et al. (2003). "Modelling air quality in street
canyons: a review." Atmospheric Environment 37(2): 155-182.
Vehkamaki, H., M. Kulmala, et al. (2002). "An improved parameterization for
sulfuric acid-water nucleation rates for tropospheric and stratospheric
conditions." Journal of Geophysical Research-Atmospheres 107(D22): -.
Venkatesan, R., R. Mathiyarasu, et al. (2002). "A study of atmospheric dispersion of
radionuclides at a coastal site using a modified Gaussian model and a
mesoscale sea breeze model." Atmospheric Environment 36(18): 2933-2942.
Venkatram, A. (2003). Validation of Concentrations estimated from air dispersion
modeling for source-receptor distances of less than 100 meters. Sacramento,
California, California Air Resources Board, Research Division.
Venkatram, A., P. Karamchandani, et al. (1997). "The development of a model to
examine source-receptor relationships for visibility on the Colorado Plateau."
Journal of the Air & Waste Management Association 47(3): 286-301.
Venkatram, A. and J. Pleim (1999). "The electrical analogy does not apply to
modeling dry deposition of particles." Atmospheric Environment 33(18):
3075-3076.
Vignati, E., R. Berkowicz, et al. (1999). "Transformation of size distributions of
emitted particles in streets." The Science of the Total Environment 235: 37-49.
Villasenor, R., C. Claiborn, et al. (2001). "Mesoscale modeling of wintertime
particulate matter episodes in eastern Washington, USA." Atmospheric
Environment 35(36): 6479-6491.
Villasenor, R., M. T. Lopez-Villegas, et al. (2003). "A mesoscale modeling study of
wind blown dust on the Mexico City Basin." Atmospheric Environment
37(18): 2451-2462.

63

Villasenor, R., M. Magdaleno, et al. (2003). "An air quality emission inventory of
offshore operations for the exploration and production of petroleum by the
Mexican oil industry." Atmospheric Environment 37(26): 3713-3729.
Wesely, M. L. (1989). "Parameterization of surface resistance to gaseous dry
deposition in regional scale numerical models." Atmospheric Environment 23:
1293-1304.
Wesely, M. L. and B. B. Hicks (1977). "Some factors that affect the deposition rates
of sulfur dioxide and similar gases on vegetation." Journal of Air Pollution
Control Association 27: 1110-1116.
Wexler, A. S., F. W. Lurmann, et al. (1994). "Modeling Urban and Regional Aerosols
.1. Model Development." Atmospheric Environment 28(3): 531-546.
Willis, G. E. and J. W. Deardorff (1981). "A Laboratory study of dispersion in the
middle of the convectively mixed layer." Atmospheric Environment 15: 109117.
Yamartino, R. J., J. S. Scire, et al. (1989). CALGRID: A Mesoscale Photochemical
Grid Model. Volume I: Model Formulation Document. Sacramento, CA,
California Air Resources Board.
Zhang, Y., B. Pun, et al. (2004). "Development and application of the model of
aerosol dynamics, reaction, ionization, and dissolution (MADRID)." Journal
of Geophysical Research-Atmospheres 109(D1): -.
Zhu, Y., W. C. Hinds, et al. (2002b). "Study of ultrafine particles near a major
highway with heavy-duty diesel traffic." Atmospheric Environment 36: 43234335.
Zhu, Y., W. C. Hinds, et al. (2002a). "Concentration and Size Distribution of
Ultrafine Particles Near a Major Highway." Journal of Air and Waste
Management Association 52: 1032-1042.

64