5 particulate matter - pm10 and pm2.5 situation in … particulate matter 5-3 table 5.1. overview on...

5 Particulate Matter 5-1

5 Particulate Matter - PM10 and PM2.5 Situation in Cyprus 5.1 Sampling Programme 5.1.1 Selection of sampling points 5.1.1.1 General aspects Ambient air quality measurements are carried out in order to achieve specific objectives with regard to stipulated air quality characteristics within a given area or at specified locations. The measurements are to be planned in such a way that the results with limit values or other definitions, with effect criteria or effect findings and with the measured values from other areas or at other times are able to be compared (Lahmann, 1997). The temporal and spatial arrangements are specified by measurements or sampling. For the planning of air pollution measurements different sampling points are selected in accordance with the VDI guideline 4280 sheet 1 (1996) as well as the Daughter Directive 1999/30/EC of the EC council. In the appendix VI of the Council Directive 1999/30/EC some specific criteria for the situation of the sampling points are stated which are briefly summarized as follows: Macroscale siting criteria Sampling points directed at the protection of human health should be sited:

� to provide data on the areas within zones and agglomerations where the highest concentrations occur to which the population is likely to be directly or indirectly exposed for a period which is significant in relation to the averaging period of the limit value(s)

� to provide data on levels in other areas within the zones and agglomerations which are representative of the exposure of the general population � A sampling point should be sited to be representative of air quality in a surrounding area

of no less than 200 m² at traffic-orientated sites and of several square kilometres at urban background sites

� Sampling points should also, where possible, be representative of similar locations not in their immediate vicinity

Sampling points directed at the protection of ecosystems or vegetation should be sited: � Distance of the sampling point of more than 20 km from agglomerations or more than 5

km from other built-up areas, industrial installations or motorways � Representative of air quality in a surrounding area of at least 1000 km²

A sampling point to be sited at a lower distance or to be representative of air quality in a less extended area, taking account of geographical conditions.

Microscale siting criteria The following guidelines should be met as far as practicable: (Appendix VI of Council Directive 1999/30/EC)

� the flow around the inlet sampling probe should be unrestricted without any obstructions affecting the airflow in the vicinity of the sampler (normally some metres away from buildings, balconies, trees, and other obstacles and at least 0,5 m from the nearest building in the case of sampling points representing air quality at the building line)

� The inlet sampling point should be between 1,5 m (the breathing zone) and 4 m above the ground also for the avoidance of the influences of sedimentation soil dust (Wurster, 1998)


� the inlet probe should not be positioned in the immediate vicinity of sources in order to avoid the direct intake of emissions unmixed with ambient air

� the sampler’s exhaust outlet should be positioned so that recirculation of exhaust air to the sampler inlet is avoided

� for all pollutants, such sampling points should be at least 25 m from the edge of major junctions and at least 4 m from the centre of the nearest traffic lane

� inlets should be sited so as to be representative of air quality near to the building line � The power supply for the sampling device (230V, 50/60 Hz, ca. 240VA) has to be

available � The measurement site should be representative for similar locations and therefore not

directly placed in the sphere of influence of a source

5.1.1.2 Selected sampling points in Cyprus In order to get the valuable information about the PM situation in Cyprus, 29 sites with different characteristics are selected for PM10 measurement. Additional PM2.5 measurements are carried out at 8 specific selected sites distributed over the whole Cyprus. As an additional objective, the locations of the measurement sites are chosen in a way that allows to identify specific sources, e.g. urban areas (domestic heating, traffic), rural areas (natural sources), and coastal areas (sea salt). A total number of 3680 24h samples is collected from different sampling sites. This total number comprises the following:

• Active sampling of particulate matter (PM) in GCC (Greek Cypriot Community):

A total of 12 representative sites for active sampling of particulate matter (PM) are selected among the primary polluted areas within the cities and their vicinity. Ten sites are located in rural areas in order to determine the natural sources of particulate matter.

• Active sampling of particulate matter (PM) in TCC (Turkish Cypriot Community): A total of seven sites for active sampling of particulate matter had been selected; two sites were located in rural areas

• Continuous daily samplings at two sites with 229 PM10 and 220 PM2.5 samples at traffic site and 236 in both PM10 and PM2.5 samples at rural site are collected to observe the temporal course of PM pollution and to cover 90% of the time per year for comparison with EU limit values:

� Traffic site “General Hospital” in Nicosia � Rural background site “Agia Marina”

These samplings are carried out by samplers permanent staying at the existing GCC monitoring stations by nearly daily filter changing.

� 13 sites with 80 samples per site: These sampling sites are located in the cities of Nicosia (GCC and TCC), in Limassol and Larnaka as well as in two rural areas in GCC and one in TCC.

� 14 sites with 60 samples per site: These sites are located in Paphos, Kyrenia, Famagusta, Morphu and in 8 rural areas in GCC and TCC.

Parallel measurements of PM10 and PM2.5 are carried out at 8 sites, 6 in GCC and 2 in TCC using the same logistics as for the PM10 measurements.

In the Table 5.1 details overview about the sites, site-characteristics (traffic, urban background,


Table 5.1. Overview on the PM sampling sites and number of samples and site classification

Zone PM10 PM2.5

Number of sampling

sites

Number of sampling per site

Number of sampling

sites

Number of sampling per site

Name Site Classification

TOTAL ALL 29 2540 8 1140 TOTAL GCC 22 2060 6 980

Nicosia 3 490 2 410 330 330 General Hospital Traffic 80 80 Latsia Residential 80 Athalassa Pheripheral/Recreation

Limassol 3 240 2 160 80 80 Traffic Traffic 80 80 Residential Residential 80 Urban background Urban background

Larnaca 3 240 0 0 80 Traffic Traffic 80 Pure residential Residential 80 Residential Residential/industrial

Paphos 2 120 0 0 60 Traffic Traffic 60 Residential Residential

Zygi 1 80 80 Rural (Industry) Village>700

Rural 10 890 2 410 330 330 Agia Marina Mountain without forests 80 80 Akaki Village< 700/Agricultural 60 Anageia Village< 700/Agricultural 60 Drouseia Village< 700 60 Kidasi Village< 700 60 Souni Village>700 60 Sfalangiotissas Agricultural 60 Tsakistra Mountain with forests 60 Makheras Mountain without forests 60 Sotira Agricultural

TOTAL TCC 7 480 2 160 Nicosia 2 160 1 80

80 80 Traffic Traffic 80 Residential Residential

Kyrenia 1 60 0 0 60 Residential Residential

Famagusta 1 60 0 0 60 Residential Residential

Morphou 1 60 0 0 60 Residential Residential

Rural 2 140 1 80 80 80 Myrtou Village< 700 60 Genagra Agricultural


mountain area, forest area, agricultural area etc.) as well as the number of PM10 and PM2.5 samples per site respectively are illustrated. In Figure 5.1 the spatial arrangement of all PM10 and PM2.5 sampling points in Cyprus is shown within a map.

Figure 5.1. Particulate Matter sampling sites 5.1.2 Selection of sampling equipment and filter material 5.1.2.1 Sampling equipment Gravimetric particulate matter sampling devices with LVS (Low Volume Sampler) are used for PMx (PM10 and PM2.5) sampling, together with manual samplers as well as sequential samplers (Leckel, TEOM series etc.). This is in accordance with the reference method defined in EU Directive 1999/30/EC and EN 12341 (1999). In Figure 5.2 a sequential sampler for particulate matter (PM10) is shown on a residential site in Paphos (south west of Cyprus). The pumped volume flow rate for PMx samples is 2.3 m³/h and the sampling duration is 24 hours as recommended in the EU Directive 99/30/EC.


Figure 5.2. Sequential sampler in Paphos for particulate matter (PM10) measurements 5.1.2.2 Filter material The choice of filter type is dependent on instrumentation and what type of analysis is going to be done after sampling. Possible small weight per unit area with low ordinal numbers of the material, uniformity concerning porosity and mass allocation, low blank values, as well as the constant of these characteristics at least with the same load are the conditions for the usefulness of the filter material for the X-RFA (X-ray Fluorescence Analysis). Empirical values showed that membrane filters from cellulose nitrate are particularly well suitable (Baumbach et al., 1991; Eberspächer und Schreiber, 1976; Murphy, 1984). The chosen filter material for PMx measurement is filer type AE 98 Schleicher & Schuell company, membrane filter with cellulose nitrate with a diameter of 47 mm (sampling diameter: 41 mm) and a pore size of 5 µm. It is required by EN 12341 that the filters are equilibrated at 20° C (±1) and 50% R.H. (±5), for 48 hours. This equilibration is performed before the filters are weighed previous to the sample collection and after sampling, before the filter is weighed again with the collected sample. In order to detect possible sources of contamination, a blank filter sample is taken weekly using the same procedure as for the particulate matter samples. 5.1.3 Sampling period PM samplings are carried out throughout the whole project. In general one device stays two weeks at a measurement point and afterwards it is moved to another site. Figure 5.3 depicts different sampling period of different sampling points of Cyprus throughout the whole year.


Figure 5.3. Sampling periods at the different sampling sites in Cyprus 5.2 Results of Sampling Programme 5.2.1 Limit values In the light of studies on the health impact of particulate matter (Schwartz et al., 1996; Dockery and Pope, 1996; Donaldson and MaCnee, 1999) the European Commission has included PM10 limit values in the new air quality directive (European Commission, 1999). The first European Union ‘Daughter Directive’ to be brought forward within the Framework Directive on Ambient Air Quality Assessment and Management proposes limit values for particulate matter with the aim of avoiding, preventing or reducing harmful effects on human health and/or the environment as a whole, as well as maintaining ambient air quality. The results of preliminary assessment of the measured PM concentrations in Cyprus have to be compared with the EU limit values. Table 5.2 and 5.3 give an overview over the limit values and assessment thresholds in accordance with the Annex III and V of the Council Directive 1999/30/EC.


Table 5.2. Limit values for particulate matter (Annex III) Averaging

period Limit value Margin of

tolerance Date by which limit value is to be met

Stage 1 1. 24 hour limit value for the protection of human health

24 hours 50 µg/m3 PM10 not to be exceeded more than 35 times per year

50% reducing linearly to reach 0% by 2005

1. January 2005

2. annual limit value for the protection of human health

calendar year 40 µg/m3 PM10 20% reducing linearly to reach 0% by 2005

1. January 2005

Stage 2 (Indicative limit values to be reviewed in the light of further information on health and environment effects, technical feasibility and experience in the application of Stage 1) 1. 24 hour limit value for the protection of human health

24 hours 50 µg/m3 PM10 not to be exceeded more than 7 times per year

to be derived from data and to be equivalent to the Stage 1 limit value

1. January 2010

2. annual limit value for the protection of human health

calendar year 20 µg/m3 PM10 50% reducing linearly to reach 0% by 2010

1. January 2010

Table 5.3. Upper and lower assessment thresholds for PM10 based on the indicative limit values for 1 January 2010 (Annex V) 24 hour average annual average upper assessment threshold 60% of limit value (30 µg/m³)

not to be exceeded more than 7 times per calendar year

70% of limit value (14 µg/m3)

lower assessment threshold 40% of limit value (20 µg/m³)

not to be exceeded more than 7 times per calendar year)

50% of limit value (10 µg/m3)

A possibly high level of air pollution in some parts of Cyprus, probably due to a large number of different particle sources, make the implementation of the proposed reduced air pollution a challenging task. The analyses are expected to give an indication whether EC Air Quality Directive 99/30/EC is complied. Where the limit values for PM10 laid down in Annex III are exceeded owing to concentrations of PM10 in ambient air due to natural events which results in concentrations significantly in excess of normal background levels from natural sources, Member States shall inform the Commission providing the necessary justification to demonstrate that such exceedances are due to natural events. It is also found on the contribution that anthropogenic and natural sources have the load of particulate matter in the ambient air of Cyprus. 5.2.2 PM10 concentration in Cyprus 5.2.2.1 Annual mean value Table 5.4 summarises the average PM10 concentrations with 98th percentile and the number of exceedances of a PM10 daily limit value recorded in different sites of Cyprus (without Sahara dust events and also including this special events). These values are compared with the EU limit


Table 5.4. Average PM10 levels and number of days exceeding the EU PM10 24h limit values for 2005 (Air Quality Directive 1999/30/EC)

*) no Sahara dust events during the sampling periods; **) 66 events from 386 exceedances: 17% exceedances are caused by Sahara dust events

Sample Location

Average value

without Sahara

dust µg/m³

Average value

including Sahara

dust µg/m³

98th percentile

µg/m³

No of samples

No of Exceed.

(> 50 µg/m³) without Sahara

% of exceed.without Sahara

No. of exceed.caused

by Sahara events

No of Exceed.

(> 50 µg/m³)

with Sahara

% of exceed.

with Sahara

% of exceed.caused

by Sahara

Traffic Pafos Traffic 63.3 67.8 270.0 56 33 59 2 35 63 6

Larnaka Traffic 60.0 60.0* 91.5 84 55 66 0 55* 66* 0 NicosiaTraffic TCC 42.5 49.9 105.8 83 21 25 3 24 29 13

Limassol Traffiic 41.0 42.9 82.7 84 25 30 1 26 31 4 General Hospital 40.1 44.8 142.6 255 66 26 7 73 29 10

Industry Zygi 38.7 46.1 237.2 79 18 23 2 20 25 10

Larnaka Refinery 33.5 39.0 89.3 83 3 4 3 6 7 50

Residential Kyrenia 45.5 52.2 147.3 50 12 24 3 15 30 20

Larnaka Residential 41.7 46.5 195.4 80 18 23 3 21 26 14 Nicosia Residential

TCC 38.7 38.7* 69.7 55 13 24 0 13* 24* 0

Limassol Residential 31.5 33.9 61.9 83 4 5 1 5 6 20 Famagusta 29.9 35.5 146.6 56 3 5 3 6 11 50

Latsia 28.9 30.9 95.8 84 2 2 2 4 5 50 Morphu 27.0 32.1 119.4 56 1 2 3 4 7 75

Pafos Residential 21.7 25.0 204.9 56 0 0 1 1 2 100

Urban background Limassol Urban

Background 38.0 38.0* 67.1 81 8 10 0 8* 10* 0

Archangelos 33.5 47.6 186.1 56 4 7 3 7 13 43 Trahoni 28.6 28.6* 54.3 28 4 14 0 4* 14* 0 Kornos 25.2 25.2* 54.9 42 2 5 0 2* 5* 0

Athalassa 20.1 30.4 122.3 41 2 5 4 6 15 67 Rural background

Sfalangiotissas 32.2 36.4 139.5 55 10 18 1 11 20 9 Sotira 26.7 35.3 173.1 56 3 5 4 7 13 57

Kofinou 21.7 21.7* 47.2 42 0 0 0 0 0 0 Suni 21.1 21.1* 62.2 53 2 4 0 2* 4* 0 Lefka 19.2 19.2* 41.3 56 1 2 0 1* 2* 0 Kidasi 18.0 18.0* 50.9 55 2 4 0 2* 4* 0 Akaki 16.1 27.4 152.4 70 1 1 5 6 9 83

AgiaMarina 15.8 20.0 83.2 241 7 3 5 12 5 42 Macheras 15.8 15.8* 40.7 57 2 4 0 2* 4* 0 Tsakistra 11.9 20.7 134.3 56 1 2 5 6 11 83 Drousia 11.2 18.2 109.8 56 0 0 5 5 9 100

∑320 ∑66** ∑386 17%

5 Particulate Matter

5-9

63,3

60

42,5

41

40,1

38,7

33,5

45,5

41,7

38,7

31,5

29,9

28,9

27

21,7

38

33,5

28,6

25,2

20,1

32,2

26,7

21,7

21,1

19,2

18

16,1

15,8

15,8

11,9

11,2

0 10 20 30 40 50 60 70 80Pafos Traffic

Larnaka Traffic

NicosiaTraffic TCC

Limassol Traffiic

General Hospital

Zygi

Larnaka Refinery

Kyrenia

Larnaka Residential

Nicosia Residential TCC

Limassol Residential

Famagusta

Latsia

Morphu

Pafos Residential

Limassol Urban Background

Archangelos

Trahoni

Kornos

Athalassa

Sfalangiotissas

Sotira

Kofinou

Suni

Lefka

Kidasi

Akaki

AgiaMarina

Macheras

Tsakistra

Drousia

PM10 Concentration in µg/m³

Figure 5.4. Average PM

10 concentrations at the different sampling points in C

yprus without S

ahara dust events

67,8

60

49,9

42,9

44,8

46,1

39

52,2

46,5

38,7

33,9

35,5

30,9

32,1

25

38

47,6

28,6

25,2

30,4

36,4

35,3

21,7

21,1

19,2

18

27,4

20

15,8

20,7

18,2

0 10 20 30 40 50 60 70 80

Pafos Traffic

Larnaka Traffic

NicosiaTraffic TCC

Limassol Traffiic

General Hospital

Zygi

Larnaka Refinery

Kyrenia

Larnaka Residential

Nicosia Residential TCC

Limassol Residential

Famagusta

Latsia

Morphu

Pafos Residential

Limassol Urban Background

Archangelos

Trahoni

Kornos

Athalassa

Sfalangiotissas

Sotira

Kofinou

Suni

Lefka

Kidasi

Akaki

AgiaMarina

Macheras

Tsakistra

Drousia

PM10 Concentration in µg/m³

Figure 5.5. Average PM

10 concentrations at the different sampling points in C

yprus including Sahara dust

events

Traffic Industry

Residential

Urban

background R

ural background

Traffic Industry

Residential

Urban

background R

ural background

2005 EU annual standard





values for 2005. The average PM10 concentrations at the different sampling points classified with different characteristics sites in Cyprus is shown in the Figure 5.4 (without Sahara dust events) and Figure 5.5 (including Sahara dust events). The sites are classified in Table 5.4 as well as in the Figures 5.4 and 5.5 in site categories and in concentration levels within the categories that means highest concentrations are listed first. The classification of the annual mean concentrations is shown in Table 5.5 which is based on the annual limit value and upper and lower assessment threshold as defined in EU Directive 1999/30/EC (see Tables 5.2 and 5.3).The complete overview of the determined average PM10 concentrations for the different sites of Cyprus classified according to Table 5.5 is shown in the map of Figure 5.6 (including Sahara dust events). The map shows that there is no site below the lower assessment threshold and Drousia and Tsakistra have concentrations between the lower and the upper assessment threshold. The concentrations of all other investigated sites are found above Table 5.5. Color selection for the map showing annual mean concentration for PM10

Mean annual PM10 concentration Color

PM10 < 10 µg/m³ Green 10 µg/m³ <PM10 < 14 µg/m³ Olive 14 µg/m³ <PM10 < 20 µg/m³ Yellow 20 µg/m³ <PM10 < 30 µg/m³ Orange 30 µg/m³ <PM10 < 40 µg/m³ light red

PM10 > 40 µg/m³ Red

Figure 5.6. Average PM10 concentrations in Cyprus including special events (e.g Saharan dust) the upper assessment threshold. It is shown that 9 sites which are indicated as red zone such as General Hospital, Nicosia Traffic TCC, Kyrenia, Larnaka Residential, Limassol Traffic, Larnaka


Traffic, Archangelos, Pafos Traffic and Zygi have exceeded the annual mean limit value (40 µg/m³) for the year 2005. Again it is found that the concentrations of about 10 sites are between the range 30 µg/m³ and 40 µg/m³, some of which are close to annual limit value. 5.2.2.2 Exceedance of 24h limit value The classes in which the determined 24h value exceedances of the different sites in Cyprus shall be classified are shown in Table 5.6. The number of exceedances of the PM10 daily limit value is calculated as in percentage related to one year values (365 days). The evaluation of the determined number of exceedances of the PM10 24h limit value, calculated in percentages according to the colors identified in Table 5.6 is illustrated in a map in Figure 5.7. Table 5.6. Color selection for the map showing the exceedances of 24h limit value for PM10

Number of exceedances of PM10 24h limit value (50 µg/m³)

number in year number in percentage

Color

0 (no exceedance) 0 green 1 to 7 0,27 to 1,92 light green

8 to 19 2,2 to 5,2 yellow 20 to 29 5,5 to 7,9 orange 30 to 35 8,22 to 9,6 light red

>35 > 9,6 red From the Table 5.4 it is found that the PM10 24h limit value would be exceeded in all the stations, with the exception of the sampling point Kofino. According to EU directive, the PM10 daily limit value should not to be exceeded more than 35 times (9,6 in percentage) in a year for 2005.

Figure 5.7. Exceedances of EU PM10 daily limit values in Cyprus without special events ( Saharan dust)


From the Table 5.4 it is also found that without Sahara dust events the measurements carried out in Larnaka Traffic site would exceed at 55 days the 2005 daily limit value out of 84 sampling days which results in 66% of exceedance, this 66% are not only valid for the measurement period but also for other periods. Most of the high PM10 events are recorded in the period October-May, although several events are also recorded in July to September. The highest frequency of PM10 peak events in spring-summer and autumn are probably due to a frequency of Sahara dust events in this period. However for the other periods, the correlation of these peak levels of PM10 and gaseous pollutants points towards a dominant anthropogenic cause for most of the exceedances of the PM10 limit value. As an average, about 17% of these exceedances (66 events from 386 exceedances, see Table 5.4), depending on the monitoring sites, are recorded under Sahara dust events. In addition to the above seasonal pattern of the peak PM10 events, higher summer PM10 background levels are measured for most of the sampling points in the study period. This seasonal trend of the background concentrations may be attributed to the following factors:

� a poor soil cover that allows re-suspension of soil particles � dry soil which is not bound � the low rainfall that accounts for a low particulate scavenging potential � the intense atmospheric convective dynamics (induced by the high radiation) occurring

mainly spring and summer, favoring re-suspension � the high spring to autumn frequency of atmospheric re-circulation scenarios (Milan et al.,

1997) that accounts for the maturing of the atmospheric air masses with the subsequent enrichment of anthropogenic and natural aerosol load

Figure 5.8 and 5.9 represent the 24h average PM10 concentration in the sampling points Agia Marina and General Hospital. The measurements carried out in General Hospital exceed 66 days

0

50

100

150

200

250

01.1

1.02

01.1

2.02

01.0

1.03

01.0

2.03

01.0

3.03

01.0

4.03

01.0

5.03

01.0

6.03

01.0

7.03

01.0

8.03

01.0

9.03

Date

Con

cent

ratio

n in

µg/

m³

24h limit value

Figure 5.8. 24h average PM10 concentration in Agia Marina from 03.11.02 to 18.08.03


the 2005 daily limit value which results 26% of exceedance (without special events). From the graphs it is found that Sahara dust intrusion occurs on 30th of May, 19th of March and 6th of April in both the sampling points in Agia Marina and General Hospital. Sahara dust events are clearly visible in the Figure 5.10 which was depicted from Agia Marina and General Hospital data recorded by TEOM (Tapered Element Oscillating Microbalance). The highest concentration of PM10 peak events under this Sahara intrusion is observed over the midnight starting from 23.00 p.m. to 1.30 a.m. on 29th and 30th of May 2003.

0

50

100

150

200

250

01.1

2.02

01.0

1.03

01.0

2.03

01.0

3.03

01.0

4.03

01.0

5.03

01.0

6.03

01.0

7.03

01.0

8.03

01.0

9.03

Date

Con

cent

ratio

n in

µg/

m³

24h limit value

Figure 5.9. 24h average PM10 concentration in General Hospital from 05.12.02 to 19.08.03

0

1000

2000

3000

4000

5000

6000

7000

22:0

022

:30

23:0

023

:30

0:00

0:30

1:00

1:30

2:00

2:30

3:00

3:30

4:00

4:30

5:00

5:30

6:00

6:30

7:00

7:30

8:00

8:30

9:00

9:30

Time

Con

cent

ratio

n in

µg/

m³

0

5

10

15

20

25

30

Tem

p (°

C),

Win

d sp

eed

(m/s

)

Figure 5.10.Saharan dust events over Agia Marina and General Hospital on 29/30.05.03

PM at General Hospital

PM at Agia Marina

Temperature

Wind speed


5.2.2.3 Element concentrations in PM10

Table 5.7 summarises the mean and maximum values of PM10 components (Pb, Cd, Ni and As) in the sampling sites General Hospital, Agia Marina, Limassol Residential and Agia Marina. The filter samples are analysed by Graphite Furnace Atomic Absorption Spectrometry (GF-AAS). The values of Pb are compared to EU First Daughter Directive (1999/30/EC) whereas the values

Table 5.7. Mean and maximum values of the PM10 components (Pb, Cd, Ni and As) compared to EU limit values (Analysed by GF-AAS) a) Sampling sites: General Hospital and Agia Marina

Elements EU limit value General Hospital Agia Marina

Annual average Number of samplings

Mean Max Number of Samples

Mean Max

ng/m³ n ng/m³ ng/m³ n ng/m³ ng/m³ Pb 500* 74 123 398 44 10 182 Cd 5** 70 0,1 0,6 50 0,2 0,8 Ni 20** 70 4,8 8,8 48 3,2 15 As 6** 12 nd1) nd 10 nd nd

b) Sampling sites: Limassol Residential and Kyrenia

Elements EU limit value Limassol Residential Kyrenia

Annual average Number of samplings

Mean Max Number of Samples

Mean Max

ng/m³ n ng/m³ ng/m³ n ng/m³ ng/m³ Pb 500* 37 53 90 21 27 100 Cd 5** 37 0,29 1,22 17 0,15 0,32 Ni 20** 26 5 11 17 6 12 As 6** 10 nd nd 10 nd nd

1) nd means below detection limit * EU limit value for Pb according to EU First Daughter Directive (1999/30/EC) ** EU limit values for Cd, As and Ni according to the proposal for a “Directive of the European Parliament and of

the Council European Commission, 2003 relating to arsenic, cadmium, mercury, nickel and polycyclic aromatic hydrocarbons in ambient air” (4th Daughter Directive)

of Cd, Ni and As are compared in accordance with the proposal for a “Directive of the European Parliament and of the Council European Commission, 2003 relating to arsenic, cadmium, mercury, nickel and polycyclic aromatic hydrocarbons in ambient air” (4th Daughter Directive). It is found that the concentration of arsenic in four sampling sites is below the detection limit.


5.2.3 PM2.5 situation in Cyprus PM2.5 measurements are carried out at 8 different specific selected sites distributed over the whole Cyprus. Since the coarse fractions of PM10 (between 2.5 and 10) are mainly due to natural sources and the fine fractions (<2.5 µm) are mostly due to anthropogenic activities, e.g. combustion, metal smelting or other industrial processes with high temperatures, (Ambient Air Pollution by Particulate Matter Position Paper), PM2.5 measurements provide information in terms of PM2.5/PM10 ratio. To establish a relationship between the varying meteorological conditions and the PM2.5/PM10 ratio, quasi-continuous measurements are carried out parallel at one traffic site and at one rural background site over one year. The WHO has concluded,” fine particles (commonly measured as PM2.5) are strongly associated with mortality and other endpoints such as hospitalization for cardiopulmonary dieses” and has recommended the development of air quality guidelines for PM2.5. Simultaneously it has stated “Continuation of PM10 measurement is indicated for public health protection” and that the coarse fraction of PM10, i.e. PM10-2.5 cannot be considered innocuous. These health-related findings the Working Group of CAFE (Clean Air For Europe) has recommended the use of PM2.5 rather than PM10 as the principal metric for assessing exposure to PM. The Group further recommended that once PM2.5 limit values have come into force and have replaced the Stage 1 PM10 limit values, the Stage 2 PM10 indicative limit values in the First Daughter Directive should be reclassified as target values with the aim to continue to control the coarse fraction, PM10-2.5. Taking into account that the attainment date for a new PM2.5 limit value would not be before 2010, the Working Group concluded that such a limit value should preferably not exceed 20 µg/m³. Given that in setting a final level, the attainability of any such level would have to be taken into account, the Working Group concluded that no single PM level should be recommended at this time. Rather values within the range 12 to 20 µg/m³ should be considered (CAFE & EPAQS, September 2003) while a lower annual PM2.5 limit has been recommended by the CEN, European Committee for Standardization (20 µg/m³, to be met by the year 2005, CEN 1997). According to WHO both a short-term (24 hours) and long-term value (annual average) should be established for PM2.5. From the rather scarce information about the frequency distribution of 24 h average limit value around 35 µg/m³ (not to be exceeded more than 10% of the days of the year) seems reasonable as a starting point. The European Commission is not obliged to act on the recommendation above but will take them into account when revising the First Daughter Directive. Table 5.8 summarizes average PM2.5 values and standard deviation with 98th percentile in 8 different sites of Cyprus. The 24h average PM2.5 concentrations in sampling point Agia Marina and General Hospital are shown in the Figure 5.11 and 5.12. From Agia Marina it is found that PM2.5 levels are considerably higher in winter with respect to other seasons although several events are also recorded in April and May. It is obvious that a fraction of African Saharan dust is still present in the <2.5 µm fraction which outbreaks on 30th of May over Agia Marina. It is clearly visible that all daily average concentrations are below the CEN 24h proposed limit value (35 µg/m³) except the Sahara intrusion.


Table 5.8. Statistics of average PM2.5 levels over Cyprus without including Sahara events

Sample Location Average Value

Standard Deviation 98th Percentile No of Samples

µg/m³ µg/m³ µg/m³ General Hospital 18,53 13,33 52,35 163 Limassol Traffic 16,61 7,60 29,39 53

Nicosia Traffic TCC 13,19 7,49 27,03 52 Limassol

Residential 10,12 4,31 17,59 56 Latsia 8,42 5,28 19,42 69

Agia Marina 7,86 5,86 20,08 134 Akaki 7,80 5,43 18,14 54 Lefka 4,86 2,82 9,25 26

0102030405060708090

01.1

1.20

02

01.1

2.20

02

01.0

1.20

03

01.0

2.20

03

01.0

3.20

03

01.0

4.20

03

01.0

5.20

03

01.0

6.20

03

01.0

7.20

03

01.0

8.20

03

01.0

9.20

03Date

Con

cent

ratio

n in

µg/

m³

CEN 24h proposed limit value

Figure 5.11. 24h average PM2.5 concentration in Agia Marina from 03.11.02 to 14.08.03.

010203040506070

01.1

2.20

02

01.0

1.20

03

01.0

2.20

03

01.0

3.20

03

01.0

4.20

03

01.0

5.20

03

01.0

6.20

03

01.0

7.20

03

01.0

8.20

03

01.0

9.20

03

Date

Con

cent

ratio

n in

µg/

m³

CEN 24h proposed limit value

Figure 5.12. 24h average PM2.5 concentration in General Hospital from 05.12.02 to 19.08.03


In General Hospital PM2.5 levels are also higher in winter specially in December and January whereas several events are recorded in summer and spring. It is found that some events are above the CEN 24h proposed limit value which occurs during winter specially in December and January. The highest frequency of PM2.5 peak concentration on 6th April is probably due to Sahara dust outbreaks in this period. Figure 5.13 illustrates the complete overview of determined average PM2.5 concentrations in different sampling points of Cyprus. From the graph it is clearly obvious that all average concentrations in different sampling points are below the CEN 2005 proposed annual standard (20 µg/m³).

0

10

20

30

40

50

Gen

eral

Hos

pita

l

Lim

asso

lTr

affic

Nic

osia

Traf

fic T

CC

Lim

asso

lR

esid

entia

l

Lats

ia

Agi

a M

arin

a

Aka

ki

Lefk

a

PM2.

5 co

ncen

trat

ion

(µg/

m³)

2005 CEN proposed annual standard

CEN 24h proposed limit value 2005 (35 µg/m³)

Figure 5.13. Average PM2.5 concentration in different sampling points in Cyprus

0

10

20

30

40

50

60

Gen

eral

Hos

pita

l

Lim

asso

lTr

affic

Nic

osia

Traf

fic T

CC

Lim

asso

lR

esid

entia

l

Lats

ia

Agi

aM

arin

a

Aka

ki

Lefk

a

Con

cent

ratio

n in

µg/

m³

PM10PM2.5

Traffic Residential Rural background

Figure 5.14. Comparison of averages of simultaneously measured PM2.5 & PM10 mass concentrations for different sampling points in Cyprus (only valid for comparable PM10 & PM2.5 data sets)


In the Figure 5.14 the averages of simultaneously measured PM2.5 & PM10 mass concentrations for different sampling points in Cyprus are compared (based on comparable data in Table 5.9). Monitoring data indicate that PM2.5 contributes on average around one thirds of the PM10 concentrations. The determination of the PM2.5/PM10 ratios may assist in distinguishing pollution events from high crustal PM episodes. Large differences between simultaneous measures of PM2.5 and PM10 are mainly attributed to crustal and marine contributions (Querol et al., 2001). Figure 5.15 shows averages of simultaneously measured PM10 vs. PM2.5 mass concentrations for 8 sampling points in Cyprus. Figure indicates that PM10 and PM2.5 mass concentrations are clearly correlated with a mean PM2.5/PM10 ratio = 0.39, demonstrating that on average the fine fraction is about 35% of the PM10 fraction.

y = 0,34xR2 = 0,42

0

10

20

30

40

50

60

70

0 20 40 60 80 100 120 140 160 180

PM10 (µg/m³)

PM2.

5 (µ

g/m

³)

Figure 5.15. PM10 vs. PM2.5 mass concentrations for 8 sampling points in Cyprus The descriptive statistics for PM10 and PM2.5 at 8 different sampling points in Cyprus is shown in Table 5.9. A closer look to the PM2.5/PM10 ratios shows that they range from 0.27 to 0.42. For a given site the PM2.5/PM10 ratio can be fairly constant. This is due to the fact that meteorology (dispersion) is the main factor controlling PM mass concentrations. It also suggests that the intensities of fine (PM2.5) and coarse PM sources covary. However there is no “universal” (i.e. valid for all sites) ratio between PM2.5 and PM10 mass concentrations. This is expected given the different sources contributing to PM2.5 and PM10-PM2.5. Comparatively lower ratios are observed for traffic sites which suggest a large contribution of re-suspended road dust to PM10-PM2.5 fraction. Higher ratios are observed at natural and rural background sites where secondary aerosol sources, which produce fine particles, are predominant.


Table 5.9. Descriptive statistics of PM10 and PM2.5 (only for comparable data)

Location Average PM10

Average PM2.5

Coarse (PM10-PM2.5)

PM2.5/PM10 Fine/Coarse

µg/m³ µg/m³ µg/m³ General Hospital 47,85 18,53 29,32 0,39 0,63 Limassol Traffic 39,36 16,61 22,74 0,42 0,73

Nicosia Traffic TCC 43,37 13,19 30,18 0,30 0,44 Limassol Residential 32,53 10,12 22,41 0,31 0,45

Latsia 31,36 8,42 22,94 0,27 0,37 Agia Marina 20,95 7,86 13,09 0,38 0,60

Akaki 21,44 7,80 13,64 0,36 0,57 Lefka 17,84 4,86 12,98 0,27 0,37

Figure 5.16 shows the PM2.5/PM10 and fine/coarse ratio for different sites, further categorized according to the levels of PM10 mass concentration. There are some clear difference in the observed PM2.5/PM10 and fine/coarse ratios. These both indicate that the coarse fraction makes a higher proportion of PM10 at the traffic site compared to the residential and rural background. This is possibly due to the increases coarse fraction at the traffic sites due to resuspension of wear dust caused by the mechanical turbulence of passing vehicles.

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

Gen

eral

Hos

pita

l

Lim

asso

lTr

affic

Nic

osia

Traf

fic T

CC

Lim

asso

lR

esid

entia

l

Lats

ia

Agi

a M

arin

a

Akak

i

Lefk

aRat

io P

M2.

5/PM

10 &

Fin

e/C

oars

e

PM2.5/PM10Fine/Coarse

Figure 5.16. PM2.5/PM10 and fine/coarse ratios for different sampling points in Cyprus It is found that superimposed on mean annual trends, sporadic natural PM inputs, such as Saharan air mass intrusion events, can significantly reduce PM2.5/PM10 ratios as a mean daily value (Querol et al., 2001), which is also observed in Cyprus as recorded in several days in March, April and May.


5.3 Filter characteristics of different sites 5.3.1 Filters in GCC and TCC The concentrations of the trace elements (Cr, Cu ,Al, K, Si, Na, Mg, Mn, Fe, Pb, Ba, S, Zn, Ti, Ca) are compared and contrasted for groups of samples and data matched to the three basic colors of aerosol filters (brown, grey or white). Chemical composition is related to filter color, that is, the brown and grey samples correspond to mineral dust and absorbing aerosol from various pollution sources, respectively, although these substances often are mixed in varying proportions. The white filters are best regarded as indicating the absence or low concentrations of dust or other absorbing aerosols, and they are often indicative of clean marine air. Relationships among atmospheric substances differ significantly in the color-stratified data subsets, and grouping the samples by color provides unique insight into the relationships among mineral dust, pollution aerosol, and other substances over the North Atlantic (Tomza et al., 2001). The Figure 5.17 shows the variations of filter color at different time period in sampling points Agia Marina and Kyrenia. Agia Marina is situated on the outskirts of a town indicated as rural background. The filter loading becomes gradually brighter and changes into brownish color which corresponds the measuring point approximately of surrounding mineral soil dust. Thus, the increasing influences of natural emissions are expected with soil resuspension process of PM10 concentration of the dust in the air. The highest concentration of the top right filter is an exception which exhibits a yellow color. This highest concentration of PM10 peak event in early Figure 5.17. Variation of Filter color in sampling points Agia Marina and Kyrenia

Agia Marina Low Concentration: 0,54 µg/m³ Date : 31.03.2003

Agia Marina Medium Concentration: 30,02 µg/m³ Date : 13.06.2003

Agia Marina High Concentration: 684,37 µg/m³

Date : 31.05.2003

Kyrenia Low Concentration: 23,59 µg/m³ Date : 13.11.2003

Kyrenia Medium Concentration: 82,55 µg/m³ Date : 15.11.2003

Kyrenia High Concentration: 165,30 µg/m³

Date :05.04.2003


summer is probably due to a frequency of Sahara dust events in this period. In Kyrenia residential clear fluctuations of filter colors are visible which indicate the dominating sources of emissions. The low and medium concentrations show gray color whereas the highest concentration shows dark brown to black color which indicates high soot portion. These grey and well defined dark color are due to the influence of the traffic and other anthropogenic sources of emissions and also a little bit of resuspended soil. Figure 5.18 depicts the variation of filter color in sampling points Limassol Traffic, Limassol residential and Larnaka residential. The loaded filters with both low and high concentration in Limassol Traffic contain dark colors which clearly indicate the probable traffic sources including gasoline and diesel vehicle soot emissions. Filters with low concentrations from Limassol Residential and Larnaka Residential show light to deep grey color which corresponds to suspended mineral dust and various other pollution sources. Again filters with high concentrations clearly represent dark colors which refer to possible sources of traffic with street dust. Figure 5.18. Variation of Filter color in sampling points Limassol Traffic, Limassol Residential and Larnaka Residential In Figure 5.19 loaded filters from General Hospital, Nicosia Traffic TCC and Nicosia Residential TCC at different time periods are shown. Due to the situation of the sampling point General Hospital, in the city centre of Nicosia close to a strongly traffic road crossing, must be counted as the highest anthropogenic sources of loading. Filter with medium concentration depicts light grey

Limassol Traffic Low Concentration: 5,26 µg/m³ Date : 29.04.2003

Limassol Traffic High Concentration: 65,13 µg/m³ Date : 26.06.2003

Limassol Residential Low Concentration: 4,72 µg/m³ Date : 21.04.2003

Limassol Residential High Concentration: 68,94 µg/m³ Date : 15.05.2003

Larnaka Residential Low Concentration: 4,90 µg/m³

Date : 28.05.2003

Larnaka Residential Medium Concentration: 68,18 µg/m³

Date : 8.11.2003


color where as high concentration filter depicts deep grey color which is owing to the influence of the traffic and other possible sources of burning, combustion and industrial sources. From Nicosia Traffic TCC and Nicosia Residential TCC it is found that even at the same measuring point and comparatively same low and high concentration, strong fluctuations of filter colors are observed which characterize the dominating sources of emission in each case. As for example in the case of Nicosia Traffic TCC, clearly deep dark color of high loaded filter (94 µg/m³) illustrates to a high proportion of soot particle where as low loading filter (15 µg/m³) shows less dark color indicating smaller influence of soot particle emitting sources. Again for the case of Nicosia Residential TCC, light grey to deep grey color filters may show the possible indication of both natural (regional natural resuspension of soil particles) and anthropogenic (road and demolition dust) mineral dust emission sources. Figure 5.19. Variation of filter color in sampling points General Hospital and Nicosia Traffic TCC and Nicosia Residential TCC 5.3.2 Filter characteristics on special events The impact of the long range transport of North African Saharan dust on PM10 levels recorded in air quality monitoring stations in Southern Europe has been demonstrated by Bergametti et al. (1989), Chester et al. (1993), Kubilay and Saydam (1995), Querol et al. (1998) and Rodriguez et al (2001). PM from this source region consists mainly of clay minerals, quartz, Ca and Mg carbonaceous particles with minor proportions of sulphate, nitrate and carbonaceous particles with major grain size modes between 1 and 25 µm (Coude-Gaussen et al., 1987; Molinaroli et al., 1993; Gilies et al., 1996, Afeti and Resch, 2000; Rodriguez et al., 2001), depending on the source area and on the transport patterns.

General Hospital Medium Concentration: 103,08 µg/m³

Date : 31.03.2003

General Hospital High Concentration: 202,10 µg/m³ Date : 31.03.2003

Nicosia Traffic TCC Low Concentration: 15,06 µg/m³ Date : 25.05.2003

Nicosia Traffic TCC High Concentration: 93,80 µg/m³ Date : 20.03.2003

Nicosia Residential TCC Low Concentration: 17,05 µg/m³ Date : 24.08.2003

Nicosia Residential TCC High Concentration: 78,21 µg/m³ Date : 05.12.2003


In Cyprus with Mediterranean basin, in addition to the local PM emissions, Saharan dust events reach on the order of 4-7 events per year, with a major frequency in the summer and winter-autumn periods. Table 5.10 summarizes PM10 concentration and mass load of Saharan dust occurring in Cyprus over the whole year. From the table it is clearly shown that highest frequency of Saharan dust occurs during 30th May 2003 in sampling point Agia Marina. Table 5.10. 24h PM10 concentration and mass load of Saharan dust from 10.10.02 to 29.08.03

Location Concentration Dust load (µg/m³) mg

Saharan Event on 20.10.2002 Larnaka Refinary 300,67 16,57 Pafos Residential 204,86 11,29

Famagusta 175,8 9,69 Zygi 417,17 22,99

Saharan Event on 19.03.2003 Sfalangiotissas 263,25 14,5

Pafos Traffic 270,01 14,88 Nicosia Traffic TCC 153,87 8,48

Latsia 160,74 8,86 Larnaka Residential 195,39 10,77

General Hospital 163,13 9 Limassol Residential 234,4 12,92

Saharan Event on 06.04.2003 General Hospital 202,1 11,15

Athalassa 166,36 9,17 Drousia 141,17 7,78 Sotira 192,01 10,58

Tsakistra 158,77 8,75 Saharan Event on 30.05.2003

Agia Marina 684,37 37,75 Limassol Traffic 198,84 10,96

Archangelos 598,44 32,98 Larnaka Residential 218,84 12,06

Akaki 407,47 22,46 Nicosia Traffic TCC 499 27,5

Figure 5.20 and 5.21 illustrate the variation of filter color due to the Saharan dust events during May 30 and March 19, 2003 .On May 30, high peaks of PM10 concentrations are recorded in five sampling points such as Agia Marina, Akaki, Limassol Traffic, Nicosia Traffic TCC and Larnaka Residential. The loaded filter of Agia Marina clearly shows the yellow color whereas filters of Akaki and Larnaka Residential also show yellow color covered with slightly greyish dust. Again the loaded filters of Limassol Traffic and Nicosia Traffic TCC show a little black color which is due to the thin layer of soot intensive color coming from traffic covered over the yellow color. On March 19, high peaks of PM10 concentrations are recorded in six sampling points such as Larnaka Residential, Latsia, Limassol Residential, General Hospital, Nicosia Traffic TCC and Pafos Traffic. The loaded filters of Larnaka Residential, Latsia and Limassol Residential show


Figure 5.20. Saharan dust outbreak during 30.05.2003 . Figure 5.21. Saharan dust outbreak during 19.03.2003 yellow color with greyish layer whereas the filters in Nicosia Traffic TCC and Pafos Traffic also show yellow color covered with soot intensive color due to traffic related sources.

Larnaka Residential Concentration: 195,39 µg/m³

Latsia Concentration: 160,74 µg/m³

Limassol Residential Concentration: 234,4 µg/m³

General Hospital Concentration: 163,13 µg/m³

Nicosia Traffic TCC Concentration: 153,87 µg/m³

Pafos Traffic Concentration: 270,01 µg/m³

Agia Marina Concentration: 684,37 µg/m³

Akaki Concentration: 407,47 µg/m³

Larnaka Residential Concentration: 218,84 µg/m³

Nicosia Traffic TCC Concentration: 499,00 µg/m³

Limassol Traffic Concentration: 198,84 µg/m³


5.4 Source Identification of PM10 The source identification analysis for PM ambient levels have been frequently based on dispersion models, in which emission inventories for various sources are used as input data to predict ambient PM concentrations. Alternative procedures have been developed based on Lagrangian trajectory models able to simulate the main processes involved in aerosol emission, formation, transport and deposition (Eldering and Cass, 1996; Kleeman and Cass, 1998). Although this procedure is consistent with the experimental data, it requires a detailed emission inventory that is not always available. Receptor modelling techniques are based on the evaluation of data acquired at receptor sites and most of them do not require previously identified emission sources (Henry et al., 1984). These types of models have played a key role in the evaluation of PM sources with respect to national air quality standards in certain countries. In the United States, the Chemical Mass Balance Model (Gertler et al., 1995; Chow et al., 1996) has been widely used, whereas in Europe receptor modelling techniques have been mainly based on methodologies that do not require chemical profiles from source emissions (Harrison et al., 1997a,b; Pio et al., 1998). Given that PM is emitted into the atmosphere by a number of anthropogenic and natural sources, the physical and chemical patterns may vary considerably. Both natural and anthropogenic emissions supply primary (direct emission of PM) and secondary (formed from gaseous precursors) PM. On a global scale (IPCC, 1996), PM emissions reach 3400 million tonnes/year. Anthropogenic sources account for only 10% of total PM emissions, whereas the natural primary PM emissions reach 85% (2900 million tonnes/year). These figures may chance drastically on a local scale in a highly traffic oriented areas in Cyprus such as General Hospital, Nicosia where large traffic emissions are mixed with urban and natural (local soil re-suspension and Saharan dust inputs) emissions. In this study, the source identification for ambient particulate samples from Cyprus is based on two methods such as Enrichment Factor (EF) and Receptor Modelling. The study focuses on the three sampling sites in Cyprus: General Hospital, Nicosia, the traffic background, Agia Marina, the rural background and Famagusta, the residential site in TCC. 5.4.1 Source identification with Enrichment Factor Enrichment Factor (EF) provides information on the extent to which trace metal concentrations in particulate matter are enriched or depleted relative to crustal or marine sources (Chester et al., 1993). Enrichment factors are calculated to indicate the extent of contribution of sources other than natural crust to the ambient elemental levels. It is a tool for deciding whether a certain element is enriched due to a certain medium. It is a double normalisation technique to present the results independent from actual concentrations. To do this, the ratio of the concentration of the element of interest to the concentration of an element specific to the medium is compared to the same ratio in a reference medium. The Enrichment Factor of an element is calculated by the following relation:

.

( / )( / )

X R PM

X R ref soil

C CEFC C

=

Where - CX and CR are the concentrations of the test element and reference element and - R is a reference element


There is no widely accepted rule for the choice of the reference element; Si, Al and Fe are usually used for this purpose. In this study, Al is used as reference element because Al is a dominant element in the soil sample found in Cyprus and it does not have significant anthropogenic sources. Therefore, the concentration of Al is not expected to change due to anthropogenic sources. In the source identification study in Cyprus, Standard Reference Material (SRM) NIST 2710 (Montana soil) is used as the reference soil in EF calculations. When calculating the EF for the particulate matter composition, there is no choice but to use an internationally accepted average soil composition such as NIST 2710 if no actual soil data is available. The need for using actual soil data, where available in EF calculations has been stated in various studies (Carrascho and Prendez, 1991). Owing to this fact, the enrichment factors for Cyprus PM calculated earlier by using NIST 2710 compilation have been recalculated by using the actual soil data obtained from analysis. Again Enrichment Factors of ambient PM are also calculated with respect to soil of surroundings of both General Hospital and Agia Marina. Table 5.11 represents the enrichment factor calculation for elements of General Hospital, Agia Marina and Famagusta. Again Figure 5.22 displays enrichment factors of ambient PM of Table 5.11. Enrichment Factor of elements for General Hospital, Agia Marina and Famagusta Enrichment Factor

Al Cr Cu K Si Na Mg Mn Fe Pb Ba S Zn Ti Ca

Enrichment factors regarding the average element concentrations in the soil in the surroundings w.r.t. General Hospital, Nicosia

1,0 4,3 14,2 3,4 0,2 5,3 2,2 1,7 0,6 2170 5,2 39,2 23 1,6 2

Enrichment factors regarding the element concentrations with respect to SRM (NIST 2710) in General Hospital

1,0 26,3 0,6 2,0 0,2 7,2 6,1 0,5 2,3 2,9 2,8 64,9 0,6 4,1 34

Enrichment factors regarding the element concentrations in the soil in direct proximity to Agia Marina

1,0 4,2 0,6 2,7 0,3 6,6 3,5 0,2 0,2 381 2,5 4,0 2,6 1,6 3,6

Enrichment factors regarding the averaged element concentrations in the soil in the further surroundings w.r.t. Agia Marina

1,0 2,1 0,8 0,8 0,4 2,8 3,0 1,5 0,5 280 5,8 40,8 9,7 1,7 1,4

Enrichment factors regarding the element concentrations with respect to SRM (NIST 2710) in Agia Marina

1,0 19,4 0,3 0,3 0,3 7,1 7,4 0,5 1,8 0,6 3,6 70,2 0,3 4,1 29

Enrichment factors regarding the averaged element concentrations in the soil in the further surroundings w.r.t. Famagusta

1,0 0,1 5,6 1,9 0,2 4,0 2,9 0,9 0,3 85,2 1,4 10,3 12,6 0,8 0,3

Enrichment factors regarding the element concentrations with respect to SRM (NIST 2710) in Famagusta

1,0 17,0 0,2 1,8 0,2 13 12 0,4 1,0 0,6 2,0 47,1 0,3 2,6 31


General Hospital with respect to soil of surroundings and reference soil (NIST 2710). In this sampling point there is no local soil data available. A clear result of the figure is that the degree of overestimation is much more significant in the PM10 fraction when different soil compositions are used. When the surrounding soil data is used as reference, the elements Si, Fe and Ti all have enrichment factors close to 1, which strengthens the hypothesis about their natural origin. But when reference soil data is used, the elements Cu, Mn and Zn show the same behaviour. Cr, K, Ti, Ca, Mn and Mg have enrichment factors below 5 when the surrounding soil data is used .It is the general practice to identify elements with enrichment factors less than 5 as “not enriched” (Parekth et al., 1989). When the reference soil data is used for enrichment calculation, elements especially Cr and Ca are observed to be highly enriched (>10). Again Na and Ba are defined as “slightly enriched” (5<EF<10) when surrounding soil data is used while Ba being “not enriched” when reference soil is used. Figure also shows that when surrounding soil data is used the elements Cu, S, Zn and Pb are found to be highly enriched indicating that these might have been contributed by anthropogenic sources. The very high EF value for Pb indicates vehicular exhausts as major sources of pollution.

0,1

1,0

10,0

100,0

1000,0

10000,0


Element

Enric

hmen

t Fac

tor

Figure 5.22. General Hospital: Enrichment Factor of ambient PM with respect to soil of surroundings and reference soil (NIST 2710) Figure 5.23 illustrates enrichment factors of ambient PM for Agia Marina with respect to local soil data, soil of surroundings and reference soil (NIST 2710). Figure shows that enrichment factors of Pb, S and Ca are overestimated when NIST 2710 is used instead of local soil as reference. when local soil data is used as reference, Cr, K, Mg, Ba, S, Ca, Zn and Ti have enrichment factors below 5 and identified as “not enriched” while Na (5<EF<10) as “slightly enriched” and Cu, Si, Mn, Fe (EF close to 1) represent as their natural origin. When surroundings soil data are used as reference, Pb and S identified as “highly enriched” while Ba and Zn defined as “slightly enriched” whereas Cr, Na, Mg, and Ti are being defined as “not enriched”. Furthermore, when NIST 2710 is used as reference soil, Cu, K, Si, Mn, Pb and Zn have enrichment factors close to 1 which indicates the association of sources of these elements to the reference Montana soil. Again in this case, Ti and Ba are defined as “not enriched” while Na and Mg are identified as “slightly enriched”.

soil of surroundings

ref.soil


0,1

1,0

10,0

100,0

1000,0


Element

Enric

hmen

t Fac

tor

Figure 5.23. Agia Marina: Enrichment Factor of ambient PM with respect to soil around the station, soil of surroundings and reference soil (NIST 2710) Figure 5.24 displays enrichment factors of ambient PM for residential point Famagusta with respect to soil of surroundings and reference soil (NIST 2710). No local soil data is available in this sampling point. When surrounding soil data is taking into account as reference, elements like Cr, K, Si, Mg, Na, Mn, Fe, Ba, Ti and Ca have enrichment factors less than 5 and indicated as “not enriched” while Cu (5<EF<10) identified as “slightly enriched” and Pb, S and Zn as “highly enriched”. Figure also shows that when reference soil (NIST 2710) data is used, the elements S, Ca, Cr, Na and Mg are found to be highly enriched and Cu, K, Si, Mn, Pb, Fe, Ba, Zn and Ti are not enriched at all.

0,1

1,0

10,0

100,0

1000,0


Element

Enric

hmen

t Fac

tor

Figure 5.24. Famagusta: Enrichment Factor of ambient PM with respect to soil of surroundings and reference soil (NIST 2710)


ref.soil

local soil


ref.soil


Therefore it can be concluded that the calculation of EF values of elements in urban and rural particulate matters need to be carried out cautiously, because the observed enrichment can be due to recycled polluted surface soil rather than the direct contribution of anthropogenic emissions to particulate matters. Since source identification techniques require good knowledge of the composition of particles emitted from the sources, the use of correct soil composition is important considering that there is at least one soil component (generally more than one) contributing to the total mass. The degree of the correctness of the apportionment result is size-dependent when literature values are used for soil composition. When PM2.5 samples are used, the apportionment study can produce quite correct results even when literature data is used instead of the actual soil composition. However when PM10 samples are used in apportionment studies, the need for actual soil composition is more significant because these samples include most of the coarse fraction particulate matters where concentrations of anthropogenic elements are determined mainly by recycled surface soil (Tuncel et al., 2001). Enrichment factors are used to find out which sources give which element composition at the different sites. Therefore this knowledge is used to identify the factors in the following multivariate statistical analysis of receptor modelling. 5.4.2 Source identification with Receptor Modelling 5.4.2.1 Receptor modelling concept Receptor modelling techniques are used in order to identify the sources of atmospheric particulate matter. There are many statistical techniques to extract information on the type of air pollution sources; factor analysis is a commonly used method for this purpose (Stevens and Pace, 1984). Factor analysis, a useful explanatory tool in multivariate statistical analysis can be applied to discover and interpret relationships among variables and to test different hypotheses. The correlation coefficient matrix and factor loading obtained from factor analysis are utilized to draw inferences about natural and anthropogenic sources of the various trace elements (Hopke et al., 1976; Sadasivan and Negi, 1990; Fergusson and Kim, 1991). There are different modifications within the factor analysis experience. The simplest form of factor analysis is Principal Component Analysis (PCA). PCA by using varimax-rotation has been used for the source identification in General Hospital and Agia Marina. This method attempts to simplify the description of a system by determining a minimum set of basis vectors that span the data space to be interpreted. It considers the structure of the variability and interdependence between random variables in a multidimensional data set (Le Maitre, 1982). The PCA procedure calculates a new set of independent variables (principal components, eigenvectors) from the original data set of independent variables (chemical species), usually from the covariance matrix (as in this study). A geometric interpretation of PCA is that it recalculates the original variable concentrations, into principal components, in decreasing order of variance (eigenvalues) so that most of the variance is described by the first few (three or four) principal components, so also reducing the dimensionality of the data set. The first eigenvector therefore represents the assemblage of chemical species which accounts for the largest proportion of the variance in the data set. It is noted that PCA remains a qualitative exploratory tool, being a descriptive technique, with no statistical model being assumed.


In order to interpret the significance of atmospheric levels of air pollutants, it is necessary to acquire some knowledge of their sources. Information about indicator variables and their connection with emissions from certain sources are found in the literature. Table 5.12 summarises the indicator variables measured in this study together with the associated sources of emission. Table 5.12. Selection of trace elements and their allocation to important emission sources of PM10

Emission source Indicator variable Literature Sea-salt Na, Cl, Mg (1), (2), (3)

Mineral dust Si, Al, Ca, Fe, Ti (2), (3), (4) Street Traffic (Exhaust emission)

� Gasoline vehicles � Diesel vehicles

Pb S

(5), (6), (7)

(8) Street Traffic (Diffusive emission)

� Tire abrasion � Brake abrasion � Road abrasion

Zn

Cu, Zn, Fe, Ba Like mineral dust

(9), (10) (9), (11)

Combustion Plants � Waste incineration

� Fuel oil firing � Biomass combustion

Cd, Zn, Cu

Ni, S K

(12), (13)

(5), (7), (14) (12), (15)

Industrial Process � Cement production � Steel production � Metal industry

Ca, Cr Fe, Mn Cd, Cu

(7), (16)

(12) (2), (16), (17)

(1) (Seinfeld and Pandis, 1997),(2) (Lee et al., 1994), (3) (Baumbach et al., 1990), (4) (Thöni et al., 1999), (5) (Baumbach, 1993), (6) (Scheff et al., 1984), (7) (Schreiber, 1984), (8) (Rogge et al., 1993), (9) ( Rauterberg-Wulff, 1998), (10) (Kumata et al., 2000), (11) (Hildmann et al., 1991), (12) (Huang et al., 1994), (13) (Dodd et al., 1991), (14) (Thurston and Spengler, 1985), (15) (Rogge et al., 1998), (16) (BUWAL, 1997b), (17) (BUWAL, 1993). 5.4.2.2 Sampling point General Hospital, Nicosia Out of 75 PM10 samples collected, four are found to be extreme outliers and are not included in the subsequent data analysis since they would strongly affect the model outcome. Thus the final data set includes 71 samples and the 17 elements (Al, Cr, Cu, K, Si, Na, Mg, Mn, Fe, Pb, S, Zn, Ti, Ca, Ni, Cd, Ba) which are detected in majority of the samples. Table 5.13 shows the PM10 average concentration for the variables measured in General Hospital. Only elements that are measured in all the samples are included in the factor analysis calculations to avoid problems with estimation of missing values. It is found that Ba is measured from low number of analysis and also there is no correlation of Ba with other elements. Therefore Ba is excluded from factor analysis calculation.


Table 5.13. Average concentrations in µg/g and descriptive statistics for PM10 Elements Mean Std.

Deviation Analysis

N Not detectable in N samples

Al 18243 11574 68 3 Cr 244 98 64 7 Cu 362 650 59 12 K 8541 3291 68 3 Si 20549 12966 68 3 Na 19697 9481 68 3 Mg 13717 8144 68 3 Mn 1336 629 65 6 Fe 18484 6300 68 3 Pb 6072 3369 68 3 S 31260 19750 68 3 Zn 882 587 68 3 Ti 2744 1551 68 3 Ca 98195 36004 68 3 Ni 235 126 68 3 Cd 6 5 68 3

Table 5.14 shows the fraction of the original variability that is explained by the 5 retained factors for the PCA calculation of PM10 mode in General Hospital. The communality measures the percent of variance of a given variable explained by all the factors and interpreted as the reliability of the indicator. If an indicator variable has a low communality, the factor model is not working well for that indicator and possibly it should be removed from the model. All communalities larger than 0.8 or 0.9 means that for this variable, all variability that has to be explained is accounted for. Communalities on the range of 0.6 to 0.7 are acceptable because some variables could appear near detection limit or have air pollution sources that could not be accounted for the PCA model. It is found that Cr has low communality less than 0.5 which means that the variability of this element is not well explained. Table 5.14. Communalities for each variable for the retained 5 factors in Principal Component Analysis (>0,75 well explained, <0,5 not explained) Elements Initial Extraction

Al 1 0,910 Cr 1 0,460 Cu 1 0,928 K 1 0,842 Si 1 0,812 Na 1 0,757 Mg 1 0,898 Mn 1 0,767 Fe 1 0,934 Pb 1 0,642 S 1 0,818 Zn 1 0,924 Ti 1 0,929 Ca 1 0,869 Ni 1 0,726 Cd 1 0,755


Table 5.15 represents the varimax rotated factor matrix for PM10 data. Only factor loadings larger than 0.10 are shown and factors having eigenvalues >1 are presented. The characteristic elements for the source types are marked in bold type. The results allow identifying five major particulate sources in General Hospital accounting for 81% of the total variance in the PM10 particle data set. The loadings are considered a correlation coefficient between the original variables and new factor variables that represent a linear combination. Loading values over 0.50 are generally considered significant although smaller ones may still provide insights into the nature of the sources. It is then necessary to interpret the pattern of elements for each factor. Table 5.15. Factor loading matrix for PM10 obtained from Principal Component Analysis with varimax rotation

Variables Factors 1 2 3 4 5

Al 0,19 0,93 Cr 0,25 0,23 0,17 0,12 0,55 Cu 0,14 0,95 K 0,44 0,78 -0,16 Si 0,33 0,79 -0,27 Na 0,14 0,75 0,34 0,13 0,21 Mg 0 0,86 0,16 0,35 Mn 0,81 0,22 0,13 0,21 Fe 0,92 0,16 0,25 Pb 0,15 0,19 -0,14 0,75 S -0,17 0,22 0,81 0,11 0,26 Zn 0,28 0,91 Ti 0,91 0,14 -0,18 -0,21 Ca 0,89 0,10 0,12 0,23 Ni 0,69 0,22 0,44 Cd 0,81 0,12 -0,29

Eigenvalues 4,97 3,04 2,32 1,57 1,08 % of variance 23,07 19,04 17,09 11,78 10,09

Figure 5.25 displays the estimated latent factors for PM10 data set with graphs coming from the Table 5.14. The first factor has high factor loading for Ca, Ti, Fe, Mn and also some association with Al, K, Si. This source type explains 23% of the total data set variance and represents local mineral dust source resuspended by traffic. The second factor accounts for 19% of the total variance and is characterised by the salt elements Na and Mg with PC loadings of 0.75 and 0.86 respectively. The factor analysis shows that a high fraction of dust components Al and Si may also be attributed to the salt components. Therefore factor 2 represents the source sea-salt with natural dust. The third factor has high loadings of different characteristic elements such as Ni, S, K, Cd and explains 17% of the total variance. S associated with Ni (a tracer element for fuel oil burning) indicates oil combustion; K and Cd probably associated with the emissions from the wood combustion and from the burning of agricultural wastes as well as high loading of S represent diffusive emission of traffic caused by diesel vehicles. Therefore factor 3 is interpreted as group of sources denoted as combustion and traffic source. The forth factor is highly associated with the


Figure 5.25. Estimated latent factors for the PM10 data record of the measuring point General Hospital, Nicosia characteristic elements Cu and Zn with high PC loading of 0.95 and 0.91 respectively. It is a clear indication that the correlation of these two elements represents the diffusive traffic source such as tire abrasion (source of Cu) and brakedrum abrasion (source of Zn) in highly dense traffic site General Hospital. This source type also accounts 12% of the total variance. The last factor still accounts for 10% of total variance. The only dominant element for this source type is Pb with a PC loading of 0.75 which clearly indicates the exhaust emission of traffic caused by gasoline vehicles. Although Cr has a minor portion of factor loading (0.55) in this factor but its influence is neglected due to its low communality. Figure 5.26 represents the PM source identification for General Hospital PM10 mode mass concentration. The first factor labelled as local mineral dust is responsible for most of the PM10 mass while sea-salt with natural dust and combustion and traffic factors account for a significant fraction of the PM10 mode mass in General Hospital. Exhaust and diffusive traffic are also associated with significant fraction of 4.0 and 4.8 µg/m³ of PM10 mass. The contribution of the rest of 19% of the total variance with 7.7 µg/m³ mass fraction is not identified here and may be reflected with secondary particles.

0,19 0,

25

0,45

0,33

0,14

0,81 0,

92

0,15

-0,1

7

0,28

0,91

0,89

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

Al Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca Ni Cd

Fact

or 1

0,93

0,23

0,79

0,75

0,86

0,22

0,16 0,19 0,22

0,14

0,0

0,2

0,4

0,6

0,8

1,0

Al Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca Ni Cd

Fact

or 2

0,17

0,14

0,78

-0,2

7

0,34

0,16

-0,1

4

0,81

-0,1

8

0,11

0,69 0,

81

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca Ni Cd

Fact

or 3

0,12

0,95

0,13

0,13

0,11

0,91

0,12 0,

22

0,12

0,0

0,2

0,4

0,6

0,8

1,0


Fact

or 4

0,55

-0,1

6

0,21 0,

35

0,21 0,25

0,75

0,26

-0,2

1

0,23

0,44

-0,2

9-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0


Fact

or 5


General Hospital Source IdentificationAverage PM10 concentration: 40.1 µg/m³

(without Sahara dust events)Factor 1

Loacl Mineral dust (resuspension by

traffic)9.2 µg/m³

Factor 2Sea-salt with natural

dust7.6 µg/m³

Factor 3Combustion and diesel

traffic6.8 µg/m³

Factor 4Traffic (Diffusive

emission)4.8 µg/m³

Factor 5Traffic (Exhaust

emission)4.0 µg/m³

Not allocated7.7 µg/m³

23%

19%17%12%10%

19%

Figure 5.26. PM source identification by Principal Component Analysis for General Hospital PM10 mass concentration without Sahara dust events 5.4.2.3 Sampling point Agia Marina In Agia Marina, 52 PM10 filter samples are collected. 5 of them are found to be highest peak concentrations due to Saharan dust events and other experimental errors and are excluded from the factor analysis calculations. Therefore 47 filter samples are included in final data set and elements excluding cases listwise are taken into account. Table 5.16 shows the PM10 average concentration for the variables measured in Agia Marina. It is found that Cu, Pb, Ba and Cd show no correlation with other elements and also have low detection limit and therefore are excluded from the analysis. Table 5.16. Average concentrations in µg/g and descriptive statistics for PM10

Elements Mean Std. Deviation

Analysis N

Al 13910 9122 31 Cr 207 107 31 K 13359 4874 31 Si 22777 20009 31 Na 17413 8415 31 Mg 11856 5722 31 Mn 1351 819 31 Fe 17220 10758 31 S 34401 17824 31 Zn 511 300 31 Ti 3615 3755 31 Ca 110029 67748 31 Ni 212 149 31


Table 5.17 shows the fraction of the original variability that is explained by the 4 retained factors for the PCA calculation of PM10 mode in Agia Marina. It is found that most of the elements are well explained by the model. Table 5.17. Communalities for each variable for the retained 4 factors in Principal Component Analysis (>0,75 well explained, <0,5 not explained) Elements Initial Extraction

Al 1 0,786 Cr 1 0,762 K 1 0,742 Si 1 0,798 Na 1 0,927 Mg 1 0,866 Mn 1 0,749 Fe 1 0,930 S 1 0,705 Zn 1 0,543 Ti 1 0,901 Ca 1 0,854 Ni 1 0,886

Table 5.18 displays the rotated PC loadings for the PM10 fractions. Four factors are obtained with eigenvalues >1 summing almost 80% of the total variance in the particle data set. Figure 5.27 represents estimated latent factor graphs for the PM10 data record which is depicted from the Table 5.17. The first factor is highly associated with the elements Fe, Si, Ti, K contributing 28% of the total variance. Also a minor fraction of Al, Mn and Ca exits in this factor. From the local soil composition it is found that Si, Fe, K, Al, Ca, Ti have high concentrations. Table 5.18. Factor loading matrix for PM10 obtained from Principal Component Analysis (PCA) with varimax rotation

Variables Factors 1 2 3 4

Al 0,39 0,45 0,65 Cr 0,86 K 0,84 0,20 Si 0,75 0,33 0,11 -0,32 Na 0,57 0,75 -0,16 Mg -0,31 0,21 0,85 Mn 0,53 0,66 0,18 Fe 0,95 0,15 S -0,47 -0,48 0,48 0,13 Zn 0,23 -0,14 0,66 0,18 Ti 0,77 0,49 -0,11 -0,23 Ca 0,23 0,80 0,39 Ni 0,94

Eigenvalues 5,13 2,89 1,36 1,07 % of variance 28% 23% 20% 9%


Figure 5.27. Estimated latent factors for the PM10 data record of the measuring Agia Marina Therefore factor 1 clearly indicates the source type local soil dust. The second factor explains 23% of the total variance and has high factor loadings of Ca, Cr, Mn, Na and also minor fractions of Al, Si, S, Ti. These elements may represent the source type soil from surrounding areas of Agia Marina. The third factor accounts for 20% of the total variance. The dominant elements of this factor type are Na and Mg with a PC loading of 0.75 and 0.86 together with small proportion of Al (0.65) and S (0.48). Therefore this factor may indicate the source type sea-salt with mineral dust. Again the PC loadings of S (0.48) and Zn (0.66) indicate that this source type is somewhat mixed with contributors from diffusive emission of traffic. The last factor to be included still explains slightly more than one unit of variance (1.07). The only dominant element for this factor is Ni with a high PC loading of 0.94 which indicates the possible oil combustion source. Figure 5.28 displays the PM source identification for Agia Marina PM10 mode mass concentration. Local soil dust dominates the highest mass fraction whereas soil from surroundings explains a significant fraction of the PM10 mass in Agia Marina. Oil combustion accounts for a small mass (1.4 µg/m³).The contribution of the rest of 20% of the total variance with 3.2 µg/m³ mass fraction is not identified here and may be influenced with secondary particles.

0,39

0,84

0,75

-0,3

1

0,53

0,95

-0,4

7

0,23

0,77

0,23

-0,6

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

Al Cr K Si Na Mg Mn Fe S Zn Ti Ca Ni

Fact

or 1

0,45

0,86

0,33

0,57

0,21

0,66

0,15

-0,4

8

-0,1

4

0,49

0,80

-0,6

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0


Fact

or 2

0,65

0,11

0,75 0,

85

0,48

0,66

-0,1

1

0,39

-0,2

0,0

0,2

0,4

0,6

0,8

1,0


Fact

or 3

0,20

-0,3

2 -0,1

6

0,18

0,13 0,

18

-0,2

3

0,94

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0


Fact

or 4


Agia Marina Source IdentificationAverage PM10 concentration: 15.8 µg/m³

(without Sahara dust events)


Factor 4Combustion source

1.4 µg/m³

Factor 3Sea-salt with mineral

dust3.2 µg/m³

Factor 2Soil from surroundings

3.6 µg/m³

Factor 1Local soil dust

4.4 µg/m³

28%

23%20%

20%9%

Figure 5.28. PM source identification by Principal Component Analysis for Agia Marina PM10 mass concentration without Sahara dust events 5.4.2.4 Sampling point Famagusta In Famagusta, 30 PM10 filter samples are collected. One of them is found to be highest peak concentrations due to Sahara dust event on October 20, 2002 and therefore is excluded from the factor analysis calculations. Table 5.19 shows the PM10 average concentration with standard deviation for the variables measured in Famagusta. From analysis it is found that Ba shows no correlation with other elements and also low detection limit and therefore is excluded from the PCA analysis. Table 5.19. Average concentrations in µg/g and descriptive statistics for PM10

Elements Mean Std. Deviation

AnalysisN

Not detectable

In N samplesAl 27615 28041 29 0 Cr 285 141 29 2 Cu 252 164 29 6 K 16471 7178 29 0 Si 21506 15260 29 0 Na 63630 58230 29 0 Mg 42843 49105 29 0 Mn 1560 908 29 1 Fe 14780 7080 29 0 Pb 1387 1059 29 1 S 48472 23606 29 0 Zn 937 430 29 0 Ti 3192 1931 29 0 Ca 165463 52614 29 0


Table 5.20 shows the fraction of the original variability that is explained by the 4 retained factors for the PCA calculation of PM10 mode in Famagusta. Most of the elements except Pb and Zn are well explained by the model. Table 5.20. Communalities for each variable for the retained 4 factors in Principal Component Analysis (>0,75 well explained, <0,5 not explained) Elements Initial Extraction

Al 1 0,868 Cr 1 0,833 Cu 1 0,814 K 1 0,870 Si 1 0,873 Na 1 0,839 Mg 1 0,936 Mn 1 0,748 Fe 1 0,860 Pb 1 0,598 S 1 0,699 Zn 1 0,513 Ti 1 0,935 Ca 1 0,833

Table 5.21 displays the varimax rotated factor loading matrix for PM10. Only factor loadings larger than 0.10 are shown and factors having eigenvalues >1 are presented. The results identify four major particulate sources in Famagusta accounting for 80% of the total variance in the PM10 particle data set. Figure 5.29 displays estimated latent factor graphs for the PM10 data record coming from the Table 5.20. Table 5.21. Factor loading matrix for PM10 obtained from Principal Component Analysis with varimax rotation

Variables

Factors 1 2 3 4

Al 0,89 -0,26 Cr 0,12 -0,14 0,83 -0,33 Cu 0,10 -0,10 0,89 K 0,73 0,54 0,21 Si 0,22 0,73 -0,50 -0,22 Na 0,29 0,84 0,20 Mg 0,97 Mn 0,74 0,10 0,43 Fe 0,90 0,20 Pb -0,28 -0,58 -0,43 S 0,17 0,76 0,21 0,21 Zn 0,63 -0,18 0,29 Ti 0,95 -0,16 Ca 0,85 0,33

Eigenvalues 5,37 2,93 1,51 1,41 % of variance 29,37 28,53 11,36 10,88


Figure 5.29. Estimated latent factors for the PM10 data record of the measuring point Famagusta The first factor presents high factor loading of Fe, Ti, Ca and also some fraction of K, Mn and Zn and contributing 29% of the total variance. Thus factor 1 is interpreted as representing local mineral dust emission source. The second factor is highly loaded on Mg, Na, Al, Si and some proportion of S and K and also explains 29% of the total variance like the first factor. The dominant elements of this factor type are Na and Mg with a PC loading of 0.84 and 0.97 together with the proportion of Al (0.89), Si (0.73), S (0.76) and K (0.54). From local soil composition it is found that Al, Si, S and K have high concentrations. Therefore, factor 2 clearly represents the source type sea-salt with soil dust. The third factor is highly associated with Cr (0.83) and also minor proportion of Si, Ca, Al which are found in high concentration in local soil data. Again considerable PC loading of Pb (-0.57) may represents the exhaust emission of traffic sources by gasoline vehicles. Thus factor 3 demonstrates combined source type traffic (gasoline) and local soil dust. The last factor explains 11% of the total variance shows high factor loading of Cu (0.89) with minor proportion of Mn, Zn, Pb, Cr. From soil composition it is found that Cu, Zn and Pb have low concentration than ambient air and Cr has high concentration in soil. Therefore, factor 4 represents natural soil dust and anthropogenic dust i.e. road dust resuspended by traffic. Figure 5.30 displays the PM source identification for Famagusta PM10 mass concentration. Local mineral dust and sea-salt with soil dust dominate the highest mass fraction (8.7 µg/m³) whereas factor 3 and 4 labelled as traffic (gasoline vehicles) with local dust and soil dust account for a significant fraction of PM10 mass in Famagusta. The contribution of the rest 20% of the total variance with 6.0 µg/m³ mass fraction is not identified here and may be influenced with secondary particles. Table 5.22 summarises the comparison of receptor model results between the sampling points in General Hospital, Famagusta and Agia Marina. From the table it is found that in General Hospital traffic is found to be the largest contributor to PM10 fraction (total contribution between 39% and about 60%; factor 3,4,5 and a part of factor 1).

0,12

0,10

0,73

0,22 0,

29

0,74

0,90

0,17

0,63

0,95

0,85

0,0

0,2

0,4

0,6

0,8

1,0

Al Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca

Fact

or 1

0,89

-0,1

4

0,54

0,73 0,

84 0,97

0,20

-0,2

8

0,76

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

Al Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca

Fact

or 2

-0,2

6

0,83

-0,1

0

-0,5

0

0,20

0,10

-0,5

8

0,21

-0,1

8

0,33

-0,8-0,6-0,4-0,20,00,20,40,60,81,0

Al Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti CaFact

or 3

-0,3

3

0,89

0,21

-0,2

2

0,43

-0,4

3

0,21 0,29

-0,1

6

-0,6-0,4-0,20,00,20,40,60,81,0

Cr Cu K Si Na Mg Mn Fe Pb S Zn Ti Ca

Fact

or 4


Famagusta Source IdentificationAverage PM10 Concentraion: 29.9 µg/m³

(without Sahara dust events)


Factor 4Local soil dust,

resuspension by traffic3.3 µg/m³

Factor 3Traffic (gasoline

vehicales) and some local resupended dust

3.3 µg/m³

Factor 2Sea-salt with soil dust

8.7 µg/m³

Factor 1Local Mineral dust

8.7 µg/m³

Figure 5.30. PM source identification by Principal Component Analysis for Famagusta PM10 mass concentration without Sahara dust events Table 5.22. Comparison of model results between General Hospital, Famagusta and Agia Marina General Hospital

Famagusta Agia Marina

Factor Sources Contribution Factor Sources Contribution Factor Sources Contribution

1

Local mineral dust,

(resuspension by traffic)

23% 1 Local mineral dust 29% 1 Local soil dust 28%

2 Sea-salt with natural dust 19% 2 Sea-salt with soil

dust 29% 2 Soil from surroundings 23%

3 Combustion and

traffic (diesel vehicles)

17% 3 Traffic (gasoline)

and local resuspended dust

11% 3 Sea-salt with mineral dust 20%

4 Traffic (Diffusive): Tire and brake

abrasion 12% 4

Local soil dust, resuspension by

traffic 11% 4 Combustion 9%

5 Traffic (Exhaust): Gasoline vehicles 10%

Total contribution

81% Total contribution 80% Total contribution 80%

Not allocated

19% Not allocated 20% Not allocated 20%

Avg. concentration without

Sahara dust

40,1 µg/m³ Avg. concentration

without Sahara dust

29,9 µg/m³ Avg. concentration

Without Sahara dust

15,8 µg/m³

Avg. concentration with

Sahara dust

44,8 µg/m³ Avg. con. With Sahara dust 35,5 µg/m³ Avg. con. With

Sahara dust 20,0 µg/m³

Sahara dust contr. 4,7 µg/m3 Sahara dust contr. 5,6µg/m3 Sahara dust contr. 4,2µg/m3

29%

29% 11% 11%

20%


Therefore the source identification calculated by Principal Component Analysis clearly shows that the dominant anthropogenic source types affecting the particulate matter composition at General Hospital are traffic induced. The traffic sources are dust resuspension by vehicles, diesel and gasoline vehicles and diffusive and exhaust emissions of traffic. Around 19% are coming from sea-salt together with natural dust and 19% cannot be explained. These considerations do not include the Sahara dust events which increase the peak and mean PM concentrations at all sites. These events have to be accounted as of natural origin. In residential site Famagusta, local mineral dust and sea-salt are the dominant sources of PM10. In rural characteristic background site Agia Marina, local soil dust and sea-salt with mineral dust source types have the largest contributor to PM10 (about 70%) can be considered as being of natural origin.

5 particulate matter - pm10 and pm2.5 situation in … particulate matter 5-3 table 5.1. overview on...

Documents