This article was downloaded by: [University of Arizona]

On: 18 July 2013, At: 07:11

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Environmental Technology
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tent20

Application and evaluation of AERMOD on the

assessment of particulate matter pollution caused by
industrial activities in the Greater Thessaloniki area
a a a
K.E. Kakosimos , M.J. Assael & A.S. Katsarou
a
Laboratory of Thermophysical Properties & Environmental Processes, Chemical
Engineering Department, Aristotle University, Thessaloniki, Greece
Published online: 09 Jun 2011.

To cite this article: K.E. Kakosimos , M.J. Assael & A.S. Katsarou (2011) Application and evaluation of AERMOD on the
assessment of particulate matter pollution caused by industrial activities in the Greater Thessaloniki area, Environmental
Technology, 32:6, 593-608, DOI: 10.1080/09593330.2010.506491

To link to this article: http://dx.doi.org/10.1080/09593330.2010.506491

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Environmental Technology
Vol. 32, No. 6, April 2011, 593–608

Application and evaluation of AERMOD on the assessment of particulate matter pollution

caused by industrial activities in the Greater Thessaloniki area
K.E. Kakosimos*, M.J. Assael and A.S. Katsarou
Laboratory of Thermophysical Properties & Environmental Processes, Chemical Engineering Department,
Aristotle University, Thessaloniki, Greece
(Received 28 January 2010; Accepted 1 July 2010)
Taylor and Francis

10.1080/09593330.2010.506491

Industrial activities are sources of high emission rates of particulate matter. The existence of many such industrial
plants close to a densely populated area can have a severe effect on human health. The effects can be even worse
when these emissions are added to existing background concentration levels. This study deals with the assessment of
the primary particulate matter pollution caused by industrial activities close to the city of Thessaloniki in Greece. An
atmospheric dispersion and regulatory model was employed, i.e. AERMOD from the US Environmental Pollution
Downloaded by [University of Arizona] at 07:11 18 July 2013

Agency. A new PM10 emission sources inventory was prepared for the main industrial plants of the area and the
annual and monthly average concentrations from 2003 to 2007 were calculated. The results from AERMOD were
validated against data from available monitoring stations and showed reasonably good agreement. It was estimated
that industry contributes approximately 30% of primary PM 10 on the western suburbs of the city and about 7% in the
city centre. The effect of the wind direction was also studied and it was illustrated that the frequent southwest winds
present higher concentration levels than the strong north ones.
Keywords: atmospheric dispersion; particulate matter; Thessaloniki; industry; AERMOD

1. Introduction heating [3–4]. The city is also very sensitive to the

Thessaloniki, the second largest city in Greece with transboundary transport of particulate matter, gaseous
more than a million inhabitants (Greater Thessaloniki pollutants and aerosols from central and eastern Europe
Area, 2001 census), is a major transportation hub for [1,5–9] as well as transport of dust from the Sahara and
the rest of southeastern Europe and hosts 20% of maritime aerosols [10–13].
Greece’s industrial activity. It is located in the north of This work focused on the industrial contribution to
Greece stretching over 20 km on a naturally amphithe- PM10 concentration levels in the region known as the
atrical site formed by low hills at the head of the Greater Thessaloniki Area (GTA) for the years from
Thermaikos Gulf (see Figure 1). In the east, Mount 2003 to 2007. GTA (see Figure 1) is an area consisting
Hortiatis forms a complex orography pattern which of the city of Thessaloniki and its adjacent zone of
favours local atmospheric circulation systems such as influence (from Sindos to Panorama and Thermi). The
sea–land breeze, valley–mountain winds, etc. as well as very high recorded PM10 concentrations in the GTA are
wind channelling phenomena. Bounded to the north by also partly attributed to the following geographical
the European continent and to the south by North reasons:
Africa, Thessaloniki is affected both by anthropogenic
and natural emission sources [1–2]. Local emissions ● The main industrial area is at Sindos in the north-
and regional pollution transport contribute to the city’s west. The strong, but not often today, northwest-
poor air quality and are reflected mostly in particular erly winds carry PM10 from the industrial area to
matter (PM) concentrations. The European Union’s the city.
24 h limit values for PM10 (50 µg·m−3) are frequently ● The most usual winds are southwest weak sea
exceeded in the urban monitoring sites and, unless breezes. They cannot clean the city’s atmosphere
drastically reduced, they will violate the limit values set by themselves, as they are weak and furthermore
for 2010 significantly. the line of tall buildings across the sea front acts
The main local anthropogenic emission sources for as a barrier to them. In addition these winds carry
shown.
Figure 1. Topographical map of GTA. Altitude isopleths are contoured at 100 m. Residential (grey) and industrial (black) areas are also shown. The locations of the stations monitoring PM 10 concentrations belonging to the network of the Central Macedonia Region (RCM) authority are also

particulate matter are traffic, industry and domestic sea salt over into the city [14].

*Corresponding author. Email: kkakosim@cheng.auth.gr

ISSN 0959-3330 print/ISSN 1479-487X online

© 2011 Taylor & Francis
DOI: 10.1080/09593330.2010.506491
http://www.informaworld.com
 594 K.E. Kakosimos et al.
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 1. Topographical map of GTA. Altitude isopleths are contoured at 100 m. Residential (grey) and industrial (black) areas
are also shown. The locations of the stations monitoring PM 10 concentrations belonging to the network of the Central Macedonia
Region (RCM) authority are also shown.

● Extensive industrial activities take place south-
west and north of the GTA, very close to the city;
there is a refinery, a cement factory and other
smaller industrial plants. Particulate matter from
these locations also affects local air quality.
● The handling of large quantities of mineral ores in
the harbour area and the surrounding quarries
(west and northeast of the city, respectively)
further increases the already high concentrations
of particulate matter in the GTA.
In order to measure and monitor concentration
levels of PM10, the authority of Central Macedonia
Region has installed four monitoring stations (all shown
in Figure 1):
(1) in the industrial area at Sindos;
(2) near the industrial area in the city suburb of
Kordelio;
(3) in the city centre in Agia Sofia Square;
(4) in the hillside suburb of Panorama.
In order to obtain a better concentration footprint, 20
passive selectors were placed during 2009 in the GTA.
The resulting PM10 measurements will be published
shortly [15].
In the past, atmospheric dispersion of particulate
matter in the city of Thessaloniki was studied by
stochastic or numerical models. For example, Slini et al.
[16] applied statistical techniques such as classification
and regression trees (CART) and neural networks (NN)
over the whole city and proved that their application is
highly related to data availability. Assael et al. [17–18]
demonstrated the lack of measurements and put forward
a methodology to successfully employ an operational
street pollution model (OSPM) in a more local, road-by-
road, approach to predicting particulate matter concen-
tration levels when insufficient data are available.
Moussiopoulos et al. [19] applied the ozone fine struc-
ture (OFIS) model over the whole city for various atmo-
spheric pollutants and found it capable of describing
observed air pollution levels satisfactorily. This study
was based on an older similar one by Tsilingiridis et al.
 Environmental Technology
[20]. Both studies used the inventory of 2001 of Sama-
ras et al. [21], where 700 industrial emissions were
employed. However, the particulate matter emission
sources were very limited and thus the dispersion
modelling depended on a large number of assumptions.
Hence in this work, a new inventory of industrial
emission sources was drawn up to improve input data for
primary particulate matter modelling. The final inven-
tory of particulate matter emission sources contains 11
small and large industrial plants. These were the ones
included in the European Pollutant Emission Register
(EPER) and also stated in Greek law (Government
Gazette 1652/B/14.8.2008) as particulate matter pollut-
ers. Three large quarries situated close to the GTA were
also added to the inventory. These 14 particulate matter
sources formed the basis of our study in order to estimate
the long-term dispersion of particulate matter in the
Downloaded by [University of Arizona] at 07:11 18 July 2013
GTA with the US Environmental Protection Agency
(USEPA) AERMOD modelling system (see below for
details of this system).
2. The dispersion model
In order to be able to calculate primary particulate
matter concentrations at various locations, dispersion
models employ mathematical equations describing the
atmospheric conditions and the dispersion, chemical
and physical processes within the plume. A review on
dispersion in street canyons [22] and comparisons of
different models using test meteorological data [23–24]
have focused on modelling gaseous dispersion.
A useful review of dispersion models and their appli-
cation to the dispersion of particles was published by
Holmes and Morawska in 2006 [25]. In this review a
large number of Box (e.g. CPB and PBM), Gaussian
(e.g. AERMOD and UK-ADMS), Lagrangian/Eulerian
(e.g. TAPM and ARIA) and CFD (e.g. MISKAM and
micro CALGRID) models were reviewed. Models
which employ special aerosol dynamics were also
included (e.g. CALGRID and CIT). The factors selected
as critical to the choice of a specific model were:
● complexity of the environment;
● dimensions of the model;
● nature of the particle source;
● computing power and time required;
● desired accuracy and timescale of the calculated
concentrations.
Hence, Gaussian models were preferred in this work
because they perform very well under steady state
conditions and for distances of a few tens of kilometres
from the source. Furthermore, they employ a large
number of various sources and they require much less
computation time for long-term studies than other types
595
of models (e.g. CFD and Lagrangian). More specifi-
cally, the Gaussian model selected was AERMOD
(version 6.2 by Lakes Environmental), which the
USEPA has adopted as its preferred regulatory model
for ranges up to 50 km. A brief description is presented
below.
2.1. The AERMOD model
AERMOD was developed for regulatory purposes in
1991 by AERMIC – American Meteorological Society
(AMS)/USEPA Regulatory Model Improvement
Committee [26]. AERMOD is a steady-state plume
model currently containing algorithms for:
● dispersion in both the convective and stable
boundary layers;
● plume rise and buoyancy;
● plume penetration into elevated inversions;
● computation of vertical profiles of wind, turbu-
lence and temperature;
● the urban night-time boundary layer;
● the treatment of receptors on all types of terrain
from the surface up to and above the plume
height;
● the treatment of building wake effects;
● an improved approach for characterizing the
fundamental boundary layer parameters;
● the treatment of plume meander.
Moreover, because each source is described in detail
in AERMOD (e.g. technical characteristics), the indus-
try can directly estimate its contribution and the poten-
tial improvement due to the application of new cleaner
technologies.
Examples of the successful use of the AERMOD
model include:
● for PM10 dispersion over Pune, India [27];
● for pollutant dispersion modelling in Hangzhou,
China [28];
● for assessment of the human health impact caused
by industrial activities in the Pace Valley of
Messina, Italy [29].
Furthermore, it has been extensively validated
against measurements and other models in various inter-
comparison studies. In the discussion by Caputo et al.
[23], it was tested against HPDM, PCCOSYMA and
HYSPLIT for modelling gaseous pollutants. Tarde and
Westbrook [30] compared it with its predecessor models
(ISCT and ISCT3) for modelling industrial PM10
sources. Again for industrial sources, Cimorelli et al.
[31] and Perry et al. [32] examined the performance
of the model against 17 field studies. Olesen et al. [33]
 596 K.E. Kakosimos et al.

validated OML, AERMOD and MISKAM using wind AERMOD employs a multiplication factor depen-
tunnel experiments for simple stack building configura- dent upon the wind speed for the correction of the emis-
tions. Sivacoumar et al. [34] compared it with FDM and sion rate in the case of open areas and quarries. These
ISCST3 for modelling fugitive dust emissions from the factors were obtained from the literature [36] for similar
stone crushing industry in Trisoolam, India. configurations and are shown in Table 2.
Finally, it should be noticed that all emission rates
were considered steady, in other words not fluctuating
3. Input data
in time. Moreover, the actual emission rates are proba-
3.1. Emission sources bly higher than the ones employed in the present study
One limiting drawback concerning the usability of since the BREFS are considered to comprise a rather
emission inventories for air quality modelling is optimistic approach.
always the accuracy of emission estimates [2,35]. A
prerequisite for compiling accurate emission invento-
ries is to assemble detailed and updated emission data 3.2. Meteorology and topography
or input parameters, which are in most cases unavail- The meteorological pre-processor AERMET uses
able or incomplete. The data frequently originate from hourly meteorological data plus surface characteris-
different sources (international, national and local tics to calculate the planetary boundary layer (PBL)
Downloaded by [University of Arizona] at 07:11 18 July 2013

authorities, public and private companies) resulting in parameters:

inconsistencies and incompatibilities, which have to be
resolved. ● friction velocity;
Past inventories for the industrial sources of partic- ● Monin–Obukhov length;
ulate matter in the GTA were based on 2–3 major indus- ● convective velocity scale;
trial activities and solely on the emissions that they ● temperature scale;
reported. In this study, for the first time, a new inven- ● mixing height;
tory was formed which includes emissions from 11 ● surface heat flux.
small and large industrial plants, and three large quar-
ries in the GTA (to our knowledge probably all existing The parameters are then passed to the main proces-
sources) due to industrial activities. These activities sor, AERMOD.
varied from processes (corresponding to pollution For this study, hourly meteorological data for the
generated during production activities), stacks and open years from 2003 to 2007 (wind speed and direction,
areas (i.e. places of PM10 fugitive dust emissions from ambient temperature and pressure, and cloud coverage)
different types of soil surfaces). All the plants and quar- were obtained by from a private weather stations
ries were asked to provide details of their PM10 emis- network [37] through the Weather Underground website
sions or sufficient data with which to calculate them (www.wunderground.com). The weather station with
by employing the recommended Integrated Pollution the code name ITHESSAL2 was selected as it is located
Prevention and Control (IPPC) calculations. Details of in the centre of the GTA and a short-term statistical
the emission sources are given in Table 1 and their loca- analysis proved that its measurements could be consid-
tion is shown in Figure 2. ered representative of the whole area.
The emission rate was calculated from the capacity
Figure 2. Map of GTA showing the position of the 14 main industrial sources examined in this study.

Figure 3 shows a typical wind rose of the prevailing

of each industrial complex (Table 1) and the specific winds in the GTA for 2005. The most usual winds for
emission factor. The emission factors for process the whole studied period (from 2003 to 2007) are south-
production activities and stacks were obtained from west sea breezes, while the strongest (but quite rare
reference documents on Best Available Technique nowadays) are north winds.
The terrain pre-processor AERMAP uses gridded
Figure 3. Typical wind rose of the prevailing winds in the GTA in 2005.

(BREFs) produced under the IPPC Directive by the

European IPPC Bureau (http://eippcb.jrc.es). Neverthe- terrain data to calculate a representative terrain height
less, the calculations give an estimate of emissions. scale for each receptor location, which is used to calcu-
In the specific case of open areas, the emission rate late the divided streamline height. The topography of
per unit area was quoted directly in the BREFs as the area, also shown in Figure 1, required by AIRMAP
equal to 6.6 × 10−6 g·s−1·m−2, and for quarries as equal was obtained from the US Geological Survey (USGS)
to 2.0 × 10−5 g·s−1m−2. Furthermore, the elevation and and through the webGIS website (www.webgis.com).
the area of process production activities and open areas The selected digital elevation model (DEM) was
(Table 1) were obtained approximately by satellite E020N90 (in GTOPO30 format), which has a resolution
pictures from Google Earth. The height and internal of 30 arc-sec (∼900 m) and is in the WGS84 global
diameter of the stack as well as gas exit temperature projection datum. For the needs of the present work the
and velocity were supplied by the plant. projection datum was changed to UTM North 39 and
 Downloaded by [University of Arizona] at 07:11 18 July 2013

Table 1. PM10 emission sources inventory for the GTA.

Source characteristics Emission 5

Capacity UTM 34N Type of Elevation Stack Stack Temperature Velocity Factor Rate Rate/area ×
Industrial source (kt·yr−1) coordinates (m) source (m) Area (m2) height (m) diameter (m) (K) (m·s−1) (kg·t−1) (g·s−1) 106 (g·s−1m−2)
TITAN Cement Co. SA 1669 1 664908 Process 163 262 × 283 – – – – 0.05 2.74 37
4507142 Open area 177 93 × 77 – – – – – 0.05 6.6
Open area 173 54 × 52 – – – – – 0.02 6.6
Open area 182 34 × 39 – – – – – 0.008 6.6
TITAN Macedonian 1230 2 664529 Process 136 99 × 227 – – – – 0.94 37.1 1700
Lime Production 4507049 Stack 135 – 40 0.7 373 17.2 0.2 7.8 –
Company Open area – 59 × 38 – – – – – 0.15 6.6
SIDENOR Steel 600 3 653526 Process 24 373 × 119 – – – – 2.76 53.3 1200
Products SA 4507624 Stack 24 – 30 4.3 343 9.6 0.01 0.2 –
Stack 24 – 21 3.2 343 6.0 0.3 6.15 –
Stack 24 – 21 3.2 343 6.9 0.3 6.15 –
Open area 28 144 × 120 – – – – – 0.11 6.6
Open area 27 157 × 120 – – – – – 0.12 6.6
Afoi D. Printzi Granitis 50 4 657490 Process 4 105 × 120 – – – – 3.0 4.7 370
4501515 – – – – – – – – − −
Open area 4 239 × 107 – – – – – 0.17 6.6
Clay brick Maliouris 200 3 648489 Process 4 123 × 119 – – – – 3.0 19.0 100
Environmental Technology

manufacturing: V. 4499964 – – – – – – – – − −
S.A. Open area 4 64 × 80 – – – – – 0.03 6.6
Brick manufacturing: 50 4 652783 Process 13 133 × 72 – – – – 1.5 2.3 240
G&D Roidis SA 4506917 Stack 13 – 5 0.3 443 142 0.5 0.79 −
Stack 13 – 5 0.3 443 142 0.5 0.79 −
Stack 13 – 9 1.2 413 8.8 0.5 0.79 −
Open area 13 76 × 72 – – – – – 0.04 6.6
(Continued)
597
 Downloaded by [University of Arizona] at 07:11 18 July 2013

Table 1. (Continued).
598

Source characteristics Emission 5

Capacity UTM 34N Type of Elevation Stack Stack Temperature Velocity Factor Rate Rate/area ×
Industrial source (kt·yr−1) coordinates (m) source (m) Area (m2) height (m) diameter (m) (K) (m·s−1) (kg·t−1) (g·s−1) 106 (g·s−1m−2)
Filkeram-Johnson SA 45 3 667818 Process 44 300 × 280 – – – – 3.0 4.37 52
4491578
Macedonian Paper Mills 90 3 643352 Process 9 351 × 154 – – – – 1.0 2.89 54
SA 4508872 Stack 9 – 30 1.8 493 8.6 0.2 0.58 –
Open area 9 97 × 85 – – – – – 0.05 6.6
Open area 7 85 × 96 – – – – – 0.05 6.6
Phosphoric Fertilizers 118 3 658792 Process 44 586 × 536 – – – – 1.35 5.03 16
Industry SA + 4505392
TOSOH Hellas AIC 200 2 653300 Process 4 252 × 202 – – – – – – 19 2
4504148 Stack 4 – 26 1.0 383 9.8 – 0.272 –
Hellenic Petroleum SA 3400 3 658730 Process 34 230 × 190 – – – – 0.0003 0.03 0.74
4504868
Process 34 137 × 241 – – – – 0.047 0.41 13
Process 34 112 × 80 – – – – 0.0003 0.001 0.1
Process 34 100 × 80 – – – – 0.1 0.24 30
TITAN Quarry 665838 Open area 211 213 × 238 – – – – – 1.01 20
4505953 Open area 237 143 × 190 – – – – – 0.54 20
K.E. Kakosimos et al.

Asvestohori Quarry 670348 Open area 468 1143 × 217 – – – – – 4.96 20

4502002 Open area 419 108 × 130 – – – – – 0.28 20
Open area 446 157 × 116 – – – – – 0.36 20
Open area 574 355 × 201 – – – – – 1.43 20
Melissohori Quarry 657547 Open area 308 251 × 435 – – – – – 2.18 20
4516565
+
Active until 2006. No data anymore on open areas.
1
EPER; 2 Personal communication; 3 Company’s website; 4 Estimated; 5 From literature or calculated unless stated differently.
 Environmental Technology 599
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 2. Map of GTA showing the position of the 14 main industrial sources examined in this study.

Table 2. Emission correction factors for open areas and main processor is ready to start. Although AERMOD
quarries [31]. calculates hourly concentration levels at each receptor,
Wind speed (m s−1) Correction factor (−) it is a regulatory model and therefore it is capable of
calculating all variations of average concentration (e.g.
0–1.54 1 annual, monthly etc.) and many statistical values such
1.54–3.09 1.25 as the percentiles. In the following sections an overview
3.09–5.14 1.67 of the results for the whole studied period and the effect
5.14–8.23 3.33 of the wind direction to the local air quality are
8.23–10.8 5.03 presented and discussed. Following this, and in order
also to evaluate AERMOD, the results are compared
with measurements of the official Central Macedonia
Region (RCM) monitoring stations.
the receptors’ grid, namely the calculation grid, was set
to 2 km × 2 km.
4.1. Annual average concentrations
The annual average industrial contribution to primary
4. Results and discussion PM10 was calculated for each one of the five years
Following introduction of all input data and the studied (2003–2007). Figure 4 shows the PM10
successful execution of the two pre-processors, the concentration contour in the GTA for 2007. Similar
 600 K.E. Kakosimos et al.
Downloaded by [University of Arizona] at 07:11 18 July 2013
Figure 3. Typical wind rose of the prevailing winds in the GTA in 2005.
contours were observed for all the studied years as the
meteorological conditions did not varied significantly.
The following points can be noted for the whole five
years’ study:
(1) Around some of the sources, the industrial
contribution to concentrations rises up to 130
µg·m−3. In the city centre, however, contribu-
tions to concentrations are only about 8 µg·m−3,
rising to 16–65 µg·m−3 in the western suburbs.
This high level of concentrations is attributed to
the high level of industrial activities present in
that area (see also Figure 2).
(2) The Hortiatis mountain on the northwest side
of the GTA forms a natural bound to the
pollution dispersion. The industrial contribu-
tion to concentrations in the western suburbs
(e.g. Panorama) is very low at less than 4
µg·m−3.
Figure 4. Annual average concentration contours of PM 10 originating from industrial sources (Table 1), 2007.
4.2. Effect of wind direction on local air quality
As shown in the wind rose diagram (Figure 3), the most
usual wind in the city is the southwest wind, while the
strong north wind (known locally as ‘Vardaris’) does
not appear often. It is of particular interest to examine
the particulate matter dispersion for these two cases:
● Figure 5 shows the effect of the usual southwest
winds (interval 9–17 May 2007, average wind
speed ∼2 m s−1). Although in the city centre
concentration levels do not seem to differ from
the annual average (see Figure 4), the whole
dispersion is shifted northeast towards the moun-
tains, where it probably gets trapped. This could
be one of the reasons for the high background
concentrations in the GTA.
● Figure 6 shows the effect of the strong north
wind (interval 11–13 January 2003, average
 Environmental Technology 601
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 4. Annual average concentration contours of PM10 originating from industrial sources (Table 1), 2007.

speed 5.7 m s−1). In this case the concentration ● traffic and domestic heating;
levels appear quite reduced on the western side of ● industry;
the GTA, but most of the city centre is still ● background.
affected by almost the same industrial contribu-
tion compared with the previous case. Though Starting with the last contribution, this includes all
the north wind transports some of the emitted PM10 resulting from natural sources (e.g. bare land, sea,
PM10 in to the city, it could improve local air forests, etc.) and transported pollution (e.g. Saharan
quality. However, its frequency of appearance is dust). The monitoring station in Panorama is considered
low and becomes even lower year by year. a suburban background station, less influenced by the
traffic and domestic heating of the city, and also far
away from the industrial area at Sindos. The PM10
Figure 5.
6. Average concentration contours of PM 10 originating from industrial sources (Table 1) for one week of 2007 with prevailing north
southwest
winds.
winds.

4.3. Comparison with measurements by the official concentration measurements of that station were
monitoring stations employed to obtain the background concentrations of
As already mentioned, the Central Macedonia Region the GTA.
authority has four stations monitoring particulate The procedure adopted was as follows: the back-
matter concentrations in the GTA. These are at ground PM10 concentration for the GTA for the time
Sindos, Kordelio, Agia Sofia and Panorama (see period concerned was obtained by subtracting the very
Figure 1). The station at Panorama was employed for small industrial contribution (about 4 µg·m−3) calcu-
the calculation of the background concentration, while lated by AERMOD for the region of Panorama from the
the other three stations were used for the validation of RCM’s measurements at the Panorama station. If one
AERMOD. adds the calculated PM10 concentrations (industrial
In general, PM10 concentrations in the GTA are contribution by AERMOD) to these background
attributed to the following emission sources/contribu- concentrations, these will therefore be equal to the
tions Manoli et al. [3]: RCM measurements minus the contribution from traffic
 602 K.E. Kakosimos et al.
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 5. Average concentration contours of PM10 originating from industrial sources (Table 1) for one week of 2007 with pre-
vailing southwest winds.

and domestic heating. The following equation describes end of the GTA, very near the industrial area, and it has
this concept: been characterized as an ‘industrial type’. Thus the
influence of traffic and domestic heating is expected to
[ RCM measurements] − [ Traffic & heating] be minimal. Indeed, in the two figures, there is reason-
able agreement between RCM measurements and the
= [ Industry (AERMOD)] + [Background ] (1)
sum of the calculated values for PM10 from AERMOD
(industrial contribution) and the background. The vali-
Another method of calculating the background PM10 dation metrics, as recommended by COST 732 [38],
is to discriminate the pollution measurements at the also supports this statement (the limit values are
suburban station (Panorama) according to the wind included in parenthesis):
direction in order to exclude pollution transported from
the local urban sources. Though this latter method (1) Fractional bias (FB) = 0.17 (|FB |<0.3)
might have been more straightforward the former was (2) Geometric mean bias (MG) = 1.17 (0.7 <MG
adopted as it was considered more consistent with the <1.3)
overall approach of AERMOD. However, the results (3) Geometric variance (VG) = 1.07 (VG <1.6)
are not expected to differ significantly; a calculation for
2004 showed that the background concentration levels A one-by-one comparison between the modelled
of the adopted method were 5–10 µg·m−3 higher than monthly averages and the measurements is also
with the alternative one. presented in Figure 9. Throughout the years of the study
The mean monthly concentrations for 2004 and 2006 (2003–2007) the winter months presented high PM
for the area of Sindos are illustrated in Figures 7 and 8. concentration levels at all the urban monitoring stations.
On average, PM10 concentrations are 40–50 µg·m−3.
2004 (annual mean concentration limit is 40 µg·m−3[38]).

This has been attributed mainly to domestic heating and

Figure 8.
7. Monthly averaged PM 10 concentrations calculated by AERMOD plus the background concentrations compared to data from the monitoring station at Sindos, 2006

The RCM Sindos monitoring station is at the far west intense traffic [11–15]. However, the monitoring station
 Environmental Technology 603
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 6. Average concentration contours of PM10 originating from industrial sources (Table 1) for one week of 2007 with pre-
vailing north winds.

Figure 7. Monthly averaged PM10 concentrations calculated by AERMOD plus the background concentrations compared to data
from the monitoring station at Sindos, 2004 (annual mean concentration limit is 40 µg·m−3[38]).
 604 K.E. Kakosimos et al.
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 8. Monthly averaged PM10 concentrations calculated by AERMOD plus the background concentrations compared to
data from the monitoring station at Sindos, 2006 (annual mean concentration limit is 40 µg·m−3[38]).

Figure 9. Comparison of modelled and measured monthly average PM 10 concentrations, 2003–2007.

 Environmental Technology 605

at Sindos is an industrial station away from the city and picture shown in Figure 12. The comparison is made
there are no data available concerning the temporal only for 2003, as after then RCM stopped reporting
variation of industrial PM emissions. These are the values of PM10 concentrations for the Agia Sofia station.
Monthly averaged PM 10 concentrations calculated by AERMOD plus the background concentrations compared to data from the monitoring station at Agia Sofia Square (annual mean concentration limit is 40 µg·m−3[38]).

reasons why we add intense local activity as a cause of From the aforementioned analysis we can conclude
Figure 12.

the differences in the concentration levels in Novem- that the industrial contribution to primary PM10 concen-
ber–December rather than simply domestic heating and tration levels in the western suburbs of GTA is about
traffic. Moreover, AERMOD does not include the 30% and in the city centre is about 7%. If, however, the
formation of secondary PM, which might be a signifi- background (about 30 µg·m−3 on average) is not taken
cant fraction of total PM – especially close to the into consideration, then the industrial contribution to
sources. The above comparison was used to success- PM10 concentration levels in the western suburbs of
fully validate AERMOD and the quality of the input GTA is about 70% and in the city centre it is 20%.
data. Moreover, the provided detailed list of the emissions’
The situation is slightly different if we move to the inventory (Table 1), along with the contour figures, can
Figure 9. Comparison of modelled and measured monthly average PM 10 concentrations, 2003–2007.

RCM Kordelio station (see Figure 1). The mean be used to identify which sources influence local air
monthly concentrations for 2004 and 2006 are illus- quality the most. The overall results of this study are in
trated in Figures 10 and 11. Note, that for the period good agreement with a previous study [19] which
from May to December 2004, and from August to employed the OFIS model to estimate the industrial
Downloaded by [University of Arizona] at 07:11 18 July 2013

September 2006, no RCM measurements were reported. contribution of various atmospheric pollutants.
In this case concentration levels are much higher, Although no actual values are given, they quote an aver-
between 60 µg·m−3− 80 µg·m−3. Furthermore, there are age of 60% for the industrial contribution to PM10
larger differences between calculated values and RCM concentration levels in the city of Thessaloniki.
measurements, which are higher during the winter
months as expected. This difference certainly could be
attributed to the increased traffic and domestic heating 5. Conclusions
during winter months. We studied the contribution of industrial activities to
2004 (annual mean concentration limit is 40 µg·m−3[38]).

The RCM monitoring station at Agia Sofia Square is PM10 concentration levels in the Greater Thessaloniki
Figure 11.
10. Monthly averaged PM 10 concentrations calculated by AERMOD plus the background concentrations compared to data from the monitoring station at Kordelio, 2006

right in the centre of the city. Here the industrial contri- Area. The study covers five years from 2003 to 2007
bution to PM10 concentration is expected to be small, and employed the atmospheric dispersion model,
while the traffic and domestic heating contribution is AERMOD. This model was selected primarily because
expected to be relatively much larger. Indeed this is the of its extended past validation on the dispersion of

Figure 10. Monthly averaged PM10 concentrations calculated by AERMOD plus the background concentrations compared to
data from the monitoring station at Kordelio, 2004 (annual mean concentration limit is 40 µg·m−3[38]).
 606 K.E. Kakosimos et al.
Downloaded by [University of Arizona] at 07:11 18 July 2013

Figure 11. Monthly averaged PM10 concentrations calculated by AERMOD plus the background concentrations compared to
data from the monitoring station at Kordelio, 2006 (annual mean concentration limit is 40 µg·m−3[38]).

Figure 12. Monthly averaged PM10 concentrations calculated by AERMOD plus the background concentrations compared to
data from the monitoring station at Agia Sofia Square, 2004 (annual mean concentration limit is 40 µg·m−3 [38]).

industrial gaseous and particle pollutants. A new emis- inventory includes 11 large and small industrial plants,
sion sources inventory for PM10 based on the European and three quarries. The required emission rates were
Pollutant Emission Register and national legislation mostly calculated from Best Available Techniques
was prepared for the local industrial activities. This Reference documents for IPPC. All other technical data
 Environmental Technology
for each source (e.g. stack height, exit temperature)
were provided by the industrial plant. The required
meteorological and topographical data were retrieved
from the Hellas–Weather network [37] and the US
Geological Survey, respectively. Measurements from
the four RCM PM10 monitoring stations were employed
for estimating the background concentration, and for
validating the input data and AERMOD.
A comparison between the calculated PM10 concen-
trations by AERMOD and the monitoring station at
Sindos (industrial type) showed reasonable agreement,
according to the calculated validation metrics [38] for
the calculated monthly averages for the years 2003–
2007; i.e. FB = 0.17 (|FB |<0.3), MG = 1.17 (0.7 <MG
<1.3) and VG = 1.07 (VG <1.6). The industrial contri-
bution to PM10 concentration levels in the western
suburbs was estimated to be 30% and in the city centre
Downloaded by [University of Arizona] at 07:11 18 July 2013
to be about 7%. The calculated monthly average
concentrations for the five-year period of this study
showed that the most affected area of the GTA is the
western suburbs because they are, in essence, between
sources.
Finally, the effect of the wind direction on local air
quality was also examined. The frequent southwest
winds contribute to the presence of higher concentration
levels than the strong north ones. Unfortunately, the
latter winds appear rarely and now even less often than
in previous years.
The conclusions and the results of this study – and
the prepared inventory – could assist local regulatory
authorities in land planning and in the preparation of
emergency plans. Moreover, each source is described in
detail as in this study (e.g. technical characteristics), the
influence/impact of each source/industrial activity can
be estimated directly and which one requires immediate
action can be identified. However, it should be noted
that there is always a significant uncertainty [2,35]
concerning the estimation of the emission rates/factors
for air quality modelling mainly because data originate
from different sources (international, national and local
authorities, public and private companies).
Acknowledgements
The authors would like to thank Professor Z. Samaras and
Associate Professor G. Tsilingiridis for providing us with
their inventory on local atmospheric pollutant emissions. We
would like also to extend our thanks to Mr A. Vavatzanidis
of the authority of Central Macedonia Region for providing
the necessary data (Ministry for the Environment, Physical
Planning and Public Works) and his valuable assistance.
References
[1] E. Katragkou, S. Kazadzis, V. Amiridis, V. Papaioannou,
[1]
S. Karathanasis, and D. Melas, PM10 regional transport
607
pathways in Thessaloniki, Greece, Atmos. Environ. 43
(2009), pp. 1079–1085.
[2] K. Markakis, A. Poupkou, D. Melas, and C. Zerefos, A
[2]
GIS based anthropogenic PM10 emission inventory for
Greece, Atmos. Pollut. Res. 1 (2010), pp. 71–81.
[3]
[3] E. Manoli, D. Voutsa, and C. Samara, Chemical char-
acterization and source identification/apportionment of
fine and coarse air particles in Thessaloniki, Greece,
Atmos. Environ. 36 (2002), pp. 949–961.
[4] C. Samara and D. Voutsa, Size distribution of airborne
[4]
particulate matter and associated heavy metals in the
roadside environment, Chemosphere 59 (2005), pp.
1197–1206.
[5] G. Kallos, M. Astitha, P. Katsafados, and C. Spyrou,
[5]
Long-range transport of anthropogenically and natu-
rally produced particulate matter in the Mediterranean
and North Atlantic: Current state of knowledge, J.
Appl. Meteorol. Clim. 46 (2007), pp. 1230–1251.
[6] D.S. Balis, V. Amiridis, C. Zerefos, E. Gerasopoulos, M.
[6]
Andreae, P. Zanis, A. Kazantzidis, S. Kazadzis, and A.
Papayannis, Raman lidar and sunphotometric measure-
ments of aerosol optical properties over Thessaloniki,
Greece during a biomass burning episode, Atmos.
Environ. 37 (2003), pp. 4529–4538.
[7] B. Rappengluck, D. Melas, and P. Fabian, Evidence
[7]
of the impact of urban plumes on remote sites in the
Eastern Mediterranean, Atmos. Environ. 37 (2003),
pp. 1853–1864.
[8]
[8] C. Zerefos, K. Ganev, K. Kourtidis, M. Tzortziou, A.
Vasaras, and E. Syrakov, On the origin of SO2 above
Northern Greece, Geophys. Res. Lett. 27 (2000),
pp. 365–368.
[9] D. Melas, C. Zerefos, S. Rapsomanikis, N. Tsangas,
[9]
and A. Alexandropoulou, The war in Kosovo: Evidence
of pollution transport in the Balkans during operation
‘Allied Force’, Environ. Sci. Pollut. Res. 7 (2000),
pp. 97–104.
[10] M.E. Koukouli, D.S. Balis, V. Amiridis, S. Kazadzis,
[10]
A. Bais, S. Nickovic, and O. Torres, Aerosol variability
over Thessaloniki using ground based remote sensing
observations and the TOMS aerosol index, Atmos.
Environ. 40 (2006), pp. 5367–5378.
[11] P. Zanis, E. Katragkou, M. Kanakidou, B.E. Psiloglou,
[11]
S. Karathanasis, M. Vrekoussis, E. Gerasopoulos, I.
Lisaridis, K. Markakis, A. Poupkou, V. Amiridis, D.
Melas, N. Mihalopoulos, and C. Zerefos, Effects on
surface atmospheric photo-oxidants over Greece during
the total solar eclipse event of 29 March 2006, Atmos.
Chem. Phys. 7 (2007), pp. 6061–6073.
[12] V. Amiridis, D.S. Balis, S. Kazadzis, A. Bais, E.
[12]
Giannakaki, A. Papayannis, and C. Zerefos, Four-year
aerosol observations with a Raman lidar at Thessaloniki,
Greece, in the framework of European Aerosol Research
Lidar Network (EARLINET), J. Geophys. Res. D: Atmos.
110 (2005), pp. 1–12.
[13] A. Papayannis, D. Balis, V. Amiridis, G. Chourdakis, G.
[13]
Tsaknakis, C. Zerefos, A.D.A. Castanho, S. Nickovic, S.
Kazadzis, and J. Grabowski, Measurements of Saharan
dust aerosols over the eastern Mediterranean using elas-
tic backscatter-Raman lidar, spectrophotometric and
satellite observations in the frame of the EARLINET
project, Atmos. Chem. Phys. 5 (2005), pp. 2065–2079.
[14] D. Voutsa, C. Samara, T. Kouimtzis, and K. Ochsenkóhn,
[14]
Elemental composition of airborne particulate matter in
the multi-impacted urban area of Thessaloniki, Greece,
Atmos. Environ. 36 (2002), pp. 4453–4462.
 608 K.E. Kakosimos et al.
[15] M.J. Assael, D. Melas, and K.E. Kakosimos, Monitor-
[15]
ing Particulate Matter Concentrations with Passive
Samplers. Application to the Greater Thessaloniki
Area, Water Air Soil Pollut. (in press) DOI: 10.1007/
S11270-009-0308.
[16] T. Slini, A. Kaprara, K. Karatzas, and N. Moussiopou-
[16]
los, PM10 forecasting for Thessaloniki, Greece,
Environ. Model. Softw. 21 (2006), pp. 559–565.
[17] M.J. Assael, M. Delaki, and K.E. Kakosimos, Local
[17]
Scale Vehicles Pollution Study in the Absence of
Sufficient Data: The Case of the City of Thessaloniki,
Proc. of Air Pollution XV, 2007, Algarve, Portugal,
invited paper.
[18] M.J. Assael, M. Delaki, and K.E. Kakosimos, Applying
[18]
the OSPM model to the calculation of PM10 concentra-
tion levels in the historical centre of the city of
Thessaloniki, Atmos. Environ. 42 (2008), pp. 65–77.
[19] Moussiopoulos, C. Vlachokostas, G. Tsilingiridis, I.
[19]
Douros, E. Hourdakis, C. Naneris, and C. Sidiropoulos,
Air quality status in Greater Thessaloniki Area and the
emission reductions needed for attaining the EU air
Downloaded by [University of Arizona] at 07:11 18 July 2013
quality legislation, Sci. Total Environ. 407 (2009),
pp. 1268–1285.
[20] G. Tsilingiridis, T. Zachariadis, and Z. Samaras,
[20]
Spatial and temporal characteristics of air pollutant
emissions in Thessaloniki, Greece: Investigation of
emission abatement measures, Sci. Total Environ. 300
(2002), pp. 99–113.
[21] Z. Samaras, G. Tsilingiridis, T. Zachariadis, I. Papageor-
[21]
giou, and E. Mimilidis, Air pollutant emissions inventory
for the urban area of Thessaloniki. Final Report to the
Organisation of the Master Plan Implementation and
Environmental Protection of Thessaloniki, Thessaloniki,
2001 (in Greek).
[22] S. Vardoulakis, B.E.A. Fisher, K. Pericleous, and N.
[21]
Gonzalez-Flesca, Modelling air quality in street canyons:
a review, Atmos. Environ. 37 (2003), pp. 155–182.
[23] M. Caputo, M. Gimenez, and M. Schlamp, Intercom-
[22]
parison of atmospheric dispersion models, Atmos.
Environ. 37 (2003), pp. 2435–2449.
[24] R. Sivacoumar, and K. Thanasekaran, Comparison
[23]
and performance evaluation of models used for vehic-
ular pollution prediction, J. Environ. Eng. 127
(2001), pp. 524–530.
[25] N.S. Holmes and L. Morawska, A review of dispersion
[24]
modelling and its application to the dispersion of parti-
cles: An overview of different dispersion models avail-
able, Atmos. Environ. 40 (2006), pp. 5902–5928.
[26] A.J. Cimorelli, S.G. Perry, A. Venkatram, J.C. Weil,
[25]
R.J. Paine, R.B. Wilson, R.F. Lee, W.D. Peters, R.W.
Brode, and J.O. Paumier, AERMOD: Description of
Model Formulation, Publication No. EPA-454/R-03-
004, US Environmental Protection Agency, Research
Triangle Park, NC, 2004.
[26]
[27] A.P. Kesarkar, M. Dalvi, A. Kaginalkar, and A. Ojha,
Coupling of the Weather Research and Forecasting
Model with AERMOD for pollutant dispersion model-
ing. A case study for PM10 dispersion over Pune, India,
Atmos. Environ. 41 (2007), pp. 1976–1988.
[28] Q. Zhang, Y. Wei, W. Tian, and K. Yang, GIS-based
[27]
emission inventories of urban scale: A case study of
Hangzhou, China, Atmos. Environ. 42 (2008), pp.
5150–5165.
[29] P. Morra, R. Lisi, G. Spadoni, and G. Maschio, The
[28]
assessment of human health impact caused by indus-
trial and civil activities in the Pace Valley of Messina,
Sci. Total Environ. 407 (2009), pp. 3712–3720.
[30] J.A. Tarde and J.A. Westbrook, Air quality modeling in
[29]
a highly industrialized valley regime: A comparison of
AERMOD-PRIME to ISCST-PRIME and ISCST3 results
for PM10 emissions, Proc. of Guideline on Air Quality
Models: The Path Forward (2003), pp. 109–119.
[31] A.J. Cimorelli, S.G. Perry, A. Venkatram, J.C. Weil, R.J.
[30]
Paine, R.B. Wilson, R.F. Lee, W.D. Peters, and R.W.
Brode, AERMOD: A dispersion model for industrial
source applications. Part I: General model formulation
and boundary layer characterization, J. Appl. Meteorol.
44 (2005), pp. 682–693.
[32] S.G. Perry, A.J. Cimorelli, R.J. Paine, R.W. Brode, J.C.
[31]
Weil, A. Venkatram, R.B. Wilson, R.F. Lee, and W.D.
Peters, AERMOD: A Dispersion model for industrial
source applications. Part II: Model performance against
17 field study databases, J. Appl. Meteorol. 44 (2005),
pp. 694–708.
[33] H.R. Olesen, R. Berkowicz, M. Ketzel, and P.
[32]
Lofstrom, Validation of OML, AERMOD/PRIME and
MISKAM using the Thompson wind-tunnel dataset for
simple stack-building configurations, Boundary Layer
Meteorol. 131 (2009), pp. 73–83.
[34] R. Sivacoumar, S. Mohan Raj, S.J. Chinnadurai, and R.
[33]
Jayabalou, Modeling of fugitive dust emission and
control measures in stone crushing industry, J. Environ.
Monit. 11 (2009), pp. 987–997.
[35] N.R. Passant, Estimation of Uncertainties in the
[34]
National Emissions Inverntory (2003), report AEAT/
ENV/R/1039t.
[36] S.K. Chaulya, M. Ahmad, R.S. Singh, L.K. Bando-
[35]
padhyay, C. Bondyopadhay, and G.C. Mondal, Valida-
tion of two air quality models for Indian mining
conditions, Environ. Monit. Assess. 82 (2003), pp. 23–43.
[37] Hellas-Weather, Network of Greek private meteorologi-
[36]
cal stations, available at http://www.meteorologia.gr/.
[38] M. Schatzmann, H. Olesen, and J. Franke, COST 732
[37]
Model Evaluation Case Studies: Approach and Results
(2010).
[39] EC, Council Directive 2008/50/EC on ambient air
[38]
quality and cleaner air for Europe, Official Journal of
the European Communities L (2008), pp. 1–44.