- Email: [email protected]
The contribution of road traffic to particulate matter and metals in air pollution in the vicinity of an urban road
Transportation Research Part D 50 (2017) 397–408
Contents lists available at ScienceDirect
Transportation Research Part D journal homepage: www.elsevier.com/locate/trd
The contribution of road traffic to particulate matter and metals in air pollution in the vicinity of an urban road Dusan Jandacka a,⇑, Daniela Durcanska a, Marek Bujdos b a b
Department of Highway Engineering, Faculty of Civil Engineering, University of Zilina, Univerzitna 8215/1, 010 26 Zilina, Slovakia Comenius University in Bratislava, Faculty of Natural Sciences, Institute of Laboratory Research on Geomaterials, Mlynska dolina, 842 15 Bratislava 4, Slovakia
a r t i c l e
i n f o
Keywords: Particulate matter (PM) Sources of PM Air pollution Metals Statistical methods
a b s t r a c t A detailed investigation was conducted to study the sources of particulate matter in the vicinity of an urban road in Zˇilina. To determine the amount of particulate matter (PM10, PM2.5 and PM1) present in the ambient air, a reference gravimetric method was used. The main objective of this contribution was to identify the sources of these particles by means of statistical methods, specifically principal component analysis (PCA), factor analysis (FA), and absolute principal component scores (APCS), as well as using the presence of 17 metals in the particulate matter (Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, Pb). To identify the metals in the particulate matter samples and to determine their abundances, spectroscopic methods were used, specifically inductively coupled plasma mass spectrometry (ICP-MS). Each of these metals may come from a specific source, such as the burning of fossil fuels in fossil fuel power plants; local heating of households; the burning of liquefied fossil fuels in the combustion engines of vehicles; the burning of coal and wood; non-combustion related emissions resulting from vehicular traffic; resuspension of traffic-related dust; and industry. Diesel vehicles and non-combustion emissions from road traffic have been identified as two key sources of the particulate matter. The results reveal that non-combustion emissions, which are associated with the elements Na, Fe, Mn, Ni, Zn, Mo, Sb, Cd, and Pb, are the major contributors, followed by combustion emissions from diesel vehicles, which are associated with the elements Mg, Ca, and Ba. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Emissions of particulate matter from different sources create a very complex mixture in the air in both qualitative and quantitative terms. Their chemical compositions (in terms of chemical elements and compounds) is the result of the distribution of all the sources in space and time and the magnitudes and characteristics of the pollutants on the one hand, and meteorological and climatic conditions on the other (Tecer, 2013; Tiwari et al., 2014; Jandacka and Durcanska, 2014; Licbinsky et al., 2010). Heavy metals are among the most basic groups of contaminants that are monitored in various parts of the environment (Balachandran et al., 2000; Chen et al., 2010). The subjects of monitoring, pursuant to the general law (Act of the National Council No. 137/2010), are the following elements: As, Cd, Hg, Pb and Ni. These are generally considered as the most harmful to people and animals. These heavy metals may prove to be highly toxic to living organisms, and at the same time they are ⇑ Corresponding author. E-mail address: [email protected] (D. Jandacka). http://dx.doi.org/10.1016/j.trd.2016.11.024 1361-9209/Ó 2016 Elsevier Ltd. All rights reserved.
398
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
very persistent in the outdoor environment (that is, they do not decompose over time through environmental processes), while at the same time, they bio-accumulate in the food chain (EEA, 2013). Some of the other elements may also be dangerous; these elements can be found in soils, where they are necessary in small amounts. However, when accumulated in large quantities, they may have a toxic impact. The following elements may have these characteristics: Cr, Co, Sn, Sb, Cu, Ni, Ag, Au, Zn, Mo, V, Mn, Fe and others (Durza, 2003). These elements (metals) are bound up with fine particles of aerosols. Heavy metals enter the environment via natural and anthropogenic processes (Weinbruch and Ebert, 2004; McCullum and Kindzierski, 2001; Gatari et al., 2006; Vojtesek et al., 2009; Thorpe and Harrison, 2008; Pant and Harrison, 2013; Sanderson et al., 2014; Kukutschova et al., 2011). Natural sources include in situ weathering processes and atmospheric deposition of metals, oceanic processes and volcanic eruptions. Anthropogenic sources include the burning of fossil fuels in order to generate electricity, raw materials excavation and ore processing, industrial processes, agricultural activities, local combustion and the continuously increasing usage of motor vehicles. In particular, vehicular traffic is the sole contributor to the presence of particulate matter in the vicinity of urban roads in medium-size cities (Chen et al., 2010). This contribution characterizes the issue of air pollution involving particulate matter in the urban environment while it attempts to identify its main sources. As the main source of pollution, we shall therefore consider vehicular traffic, which produces combustion-related emissions, non-combustion related emissions, and road dust. The profiles of metals in particulate matter was used to identify the sources in an urban area. The links between the different chemical elements (metals) were identified by using multivariate statistical methods, specifically PCA, FA, and APCS, in conjunction with Multivariate Regression Analysis (MRA) (Kachigan, 1991; Manly, 2004; Meloun and Militky, 2006; Meloun et al., 2012; Spencer, 2013; Varmuza and Filzmoser, 2009; Lu et al., 2010; Yang et al., 2011; Manta et al., 2002; Guo et al., 2004; Song et al., 2006). Based on the presence of metals in the resulting groups (factors), these groups were identified as sources of PM. 2. Method of measurements and analysis 2.1. Study area The monitoring station was situated in Zˇilina city centre and on Vojtech Spanyol Street. This street represents one of the arteries connecting the secondary and tertiary city ring roads and connects the city centre with the largest housing estate of Vlcˇince. This is reflected in the amount and frequency of traffic which passes along this city radial road. The traffic volume on certain days amounts to 15,000 vehicles/24 h. The adjacent buildings along the road include housing developments and the civil amenities and facilities. The buildings create barriers on both sides of the road. We can consider this bounded space to be a street canyon. The ventilation of the canyon is limited in two ways. There is often a large accumulation of pollution during bad weather conditions, which may include inversions and windless conditions, etc. Walkways for the pedestrians occupy both sides of the monitored road. The monitoring devices were situated close to the edge of the road. 2.2. Sample collection Sampling of particulate matter was performed near the urban connecting route during 2010 (19–25 October), 2011 (8–14 March, 11–17 April, 7–14 July, 13–19 October), and 2012 (26 January–1 February, 16–22 April, 7–13 June). The goal was long-term monitoring of a representative sample of particulate matter in the atmosphere and its behaviour relative to environmental conditions. In the second phase of the research, a chemical analysis of the particulate matter was performed, as well as the determination of its possible sources. To establish the amount of particulate matter present in the ambient air, a reference method (the gravimetric method) was applied, pursuant to the standards of STN EN 12341 (2016). The sampling was performed using low volume flow samplers (LECKEL LVS3, Low Volume Samplers). In total, 3 pieces were used. Three fractions of particulate matter were monitored concurrently, specifically PM10, PM2.5 and PM1. The coarse fraction PM2.5–10 was determined as the difference between fractions PM10 and PM2.5, and the PM1–2.5 fraction was determined as the difference between fractions PM2.5 and PM1. Particulate matter was collected on nitrocellulose filters having a diameter of 47 mm at a fixed airflow rate of 2.3 #Nm3/h (normal cubic meters per hour) under standard conditions of 101.325 kPa and 0 °C, and then the mass of particulate matter collected on the filters was determined. All filters used were conditioned at 20 °C and 48.2% RH prior to sampling, as well as after sampling and weighing (STN EN 12341, 2016). The particulate matter was sampled on the filters for 16 h during daytime sampling (6 am–10 pm) and for 8 h during night sampling (10 pm–6 am). The filters were changed in the samplers twice every 24 h (at 6 am and at 10 pm). At the end of sampling, we had 108 samples of each PM fraction eligible for further analysis. Alongside the monitoring of PM, basic meteorological data were observed as well, including temperature, ambient humidity, speed and direction of wind, and precipitation. 2.3. Sample preparation and analysis The particulate matter contains various elements and compounds. In the next phase of the research, we concentrated on the measurement of the selected metals found in the fractions PM10, PM2.5 and PM1. We focused on monitoring 17 metals
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
399
which we selected based on the assumption and probabilities that they originate in the vehicular traffic (Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, Pb). Each of these metals may come from a specific source (Weinbruch and Ebert, 2004; McCullum and Kindzierski, 2001; Gatari et al., 2006; Vojtesek et al., 2009; Thorpe and Harrison, 2008; Pant and Harrison, 2013; Sanderson et al., 2014; Kukutschova et al., 2011). To identify and to determine the amounts of chemical elements (metals) in the samples of particulate matter, spectroscopic methods were utilized (ICP-MS). The analyses of the 108 filters of each PM fraction and the determination of metals present in the fractions PM1, PM2.5 and PM10 were performed pursuant to the standard STN EN 14902 (2006). All solutions were prepared using deionized water (Type 1, Labconco WaterPro). Reagents used for sample treatment included 65% nitric acid and 40% hydrofluoric acid (all Suprapur, Merck, Germany). Calibration standards were prepared from 100 mg/l batch standard solution Astasol-MIX CZ9090 MN1 (Analytika, Czech Republic) and 1000 mg/l Sb standard solution (CertiPUR, Merck, Germany). Nitrocellulose membrane filters with the sampled airborne particulate matter were placed into 15 ml polypropylene test tubes. 2.0 ml of HNO3 and 0.67 ml of HF was added, and the tubes were sealed. Samples were left for 14 days at laboratory temperature to dissolve with one daily agitation. Subsequently, 7.33 ml of deionized water was added to each sample (the final volumes were 10 ml), and the undissolved residues were removed from the digests by centrifugation (10 min at 3000g). Working solutions were prepared by dilution of the digests and 10 ppb Rh was added as an internal standard. The measurements were performed using a quadrupole ICP-MS spectrometer (Perkin-Elmer Sciex Elan 6000). 3. Results and discussion 3.1. Particulate matter and heavy metals concentrations The overall number of measurements and the data on particulate matter concentration used represents 108 measurements for each fraction, PM1, PM2.5 and PM10. The average concentration was calculated from the day and night concentrations of PM in a given sampling period (i.e., 7 daily values and 7 night values were observed for each fraction). Maximum concentrations of PM were observed during the sampling period in January 2012 (average concentrations for 7 days PM1 = 83.8 lg/m3, PM2.5 = 112.6 lg/m3, PM10 = 125.0 lg/m3), and the minimum values were observed during the sampling period in June 2012 (average concentrations for 7 days PM1 = 12.0 lg/m3, PM2.5 = 17.2 lg/m3, PM10 = 22.4 lg/m3). The sampling period of January 2012 was accompanied by very low temperatures (the lowest daily average temperature was 11.5 °C) and frequent inversions occurred over the entire sampling period during which measurements were taken, which may to a large extent affect the PM emissions produced in terms of their accumulation in the breathing zone. On the other hand, the sampling period during June 2012 was characterized by multiple rainfall events and windy weather which resulted in the ambient air having lower PM concentrations. Moreover, during this period, conditions were very suitable for dispersion and the rainfall contributed to the natural elimination of the PM from the air by means of wet deposition. The distributions of the size fractions of PM during the various sampling periods were equally variable. The fine fraction PM2.5 had the highest share of the PM10 fraction (90%) during the sampling period in January 2012 (Fig. 1). The coarse fraction, PM2.5–10, achieved the highest fraction of PM10 during the period of April 2011, when this value reached 35% of the PM10 (Fig. 1). On average, during all the monitoring periods, the fine fraction of PM2.5 amounted to 77% by weight of PM10 (the ratio PM2.5/PM10 was 0.77, on average) (Tiwari et al., 2014). Similar results have been observed over the whole of Europe, whereas it is the fine fraction of PM2.5 which has been shown to have negative impacts on the health of the population (EEA, 2013). A number of metals are well represented in the coarse fraction (>60%), namely Na, Al, Fe, Mn, Mo, and Ba. Metals well represented in the fine fraction (>60%) include Zn, As, Cd, and Pb. The metals which are approximately equally represented in both fractions are Mg, V, Ni, Sb, Ca, Cr, and Cu (Fig. 2).
Fig. 1. Distribution of PM fractions in the total PM10 fraction during different months.
400
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 2. The average concentrations and presence of metals in the PM fractions.
3.2. Identification of particulate matter sources Multivariate statistical analyses using PCA and FA were used for statistical assessment (Kachigan, 1991; Manly, 2004; Meloun and Militky, 2006; Meloun et al., 2012; Spencer, 2013; Varmuza and Filzmoser, 2009; Lu et al., 2010; Yang et al., 2011; Manta et al., 2002; Guo et al., 2004; Song et al., 2006). The appropriateness of the use of factor analysis is proven by the KMO test (Kaiser-Meyer-Olkin) and MSA (Measure of Sampling Adequacy) (Meloun et al., 2012). To approximate the contributions from specific sources of PM, the method of APCS in conjunction with MRA (Guo et al., 2004; Song et al., 2006) is adopted. The input data for the purposes of identification of possible PM sources consisted of the concentrations of particular chemical elements, specifically metals (ng/m3), as well as the concentrations of PM (lg/m3). The concentration of particulate matter was added as a variable to the input matrix in order to detect which factor is most associated with each fraction of the particulate matter. The data matrix consisted of 18 variables and 108 items, for a total of (18 108) data objects. Multivariate statistical analyses and source identification were processed separately for the fractions PM1, PM2.5 and PM10. 3.2.1. PM1 To establish the appropriateness of the factor analysis adopted here, the KMO (interval 0–1) and MSA (interval 0–1) criteria were calculated. In accordance with these criteria, the adoption of factor analysis is substantiated. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.82, reflecting very good correlation among the variables; MSA > 0.5, indicating that the ith variable is at least moderately predicted by the others). When it is deemed to be necessary to specify the number of factors before the factor analysis calculation can be run, the analysis of principal components was performed beforehand. As a result, we could determine the number of principal components which, to a sufficient degree, account for the variability in the data. As is seen in Fig. 3, we are able to fix the number of principal components at 4 (Eigenvalue > 1). The biplot shown in Fig. 4 simplifies our decision-making process over the course of selecting the number of factors for use in factor analysis, where close relationships among several variables can be observed. In this case, there are two collections of variables separated by a significant disparity. That is, there are 2 principal components which may be consecutively used in factor analysis. Two factors were selected for factor analysis. These two principal components account for 60.5% of the total variability in the original data. The third and fourth factors were eliminated, which resulted from the consecutive factor analysis using the four factors. The fourth factor was made up primarily of one element, chromium, and that is why it represented a specific factor which is at the same time difficult to define and interpret. The third factor was composed of the elements of aluminium and vanadium, whereas following the reduction of factors; these two elements were partially moved into the second factor. In fact, following that reduction, the second factor had a more apparent form, and it became feasible to identify it. The Varimax model was used for factor rotation. The factor loadings associated with variables may be explained as the correlations between the factors and the variables. They represent the most important unit of information that the interpretation of the factors is based on. Each factor contains information about contributions of selected elements (metals). As the
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
401
Fig. 3. Graph of ‘‘foothills” eigenvalues based on PCA (PM1).
Fig. 4. Principal components biplot (PM1).
most decisive loading factors, values close to or >0.7 were selected. Based on the representation and presence of elements in particular factors, the following factors may be named (Figs. 5 and 6). The factors (pollution sources) for PM1 (Figs. 5 and 6), PM2.5 (Figs. 10 and 11) and PM10 (Figs. 15–17) were identified based on the origins of metals, the characteristics of the measuring station, the distributions of the chemical elements in the PM fractions (Fig. 2), and seasonal changes in the concentrations of metals and PM. Tyre tread, a source of airborne particles, contains natural rubber copolymers such as styrene-butadiene rubber and polyisoprene rubber, and zinc (Zn) is added to tyre tread as zinc oxide and organozinc compounds to facilitate the vulcanization process. The chemical element Zn was chosen as marker for tyre tread (Thorpe and Harrison, 2008; Pant and Harrison, 2013). A number of metals are used extensively in brake lining materials. Fe, Cu, Ba, and Sb have been commonly identified in brake
402
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 5. Factor loadings of metals on Factor 1 – F1 (PM1).
Fig. 6. Factor loadings of metals on Factor 2 – F2 (PM1).
dust samples (Thorpe and Harrison, 2008; Pant and Harrison, 2013). The chemical elements Mg and Ca were chosen as markers for exhaust sources, specifically from diesel fuel (Sanderson et al., 2014; Liati et al., 2012). Road surfaces are a site of deposition for particles from a wide variety of sources. A fraction of the PM resulting from abrasion processes will be sufficiently large to settle out under gravity and deposit on the road surface. Road dusts and roadside soils often contain significant concentrations of a range of metals, including Pb, Cu, Cd, and Zn, which are indicative of contamination by road vehicle emissions. Ions such as Ca2+, SO24 , Cl , NO3 , K+, and Na+ were also found to be present in re-suspended dust (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Han et al., 2007). Moreover, the variables (elements) with factor loadings of <0.7 do contribute to a given factor by virtue of their weight and may facilitate its naming. Factors obtained via factor analysis of PM1 were interpreted as Factor 1, representing nonexhaust traffic sources and road dust (Fig. 5), and Factor 2, representing diesel fuel (Fig. 6) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). Particulate matter PM1 is most associated with Factor 1, representing non-exhaust traffic sources and road dust. PM1 correlates most strongly with the chemical elements (metals) which characterize Factor 1. The estimated contributions of the particular factors to the creation of PM1 are given in Fig. 7 with respect to the explained original variability in the data by the three factors, amounting to the value of 60%. Factor 1, representing nonexhaust traffic sources and road dust, contributes 80.9% to the creation of PM1 and Factor 2, representing diesel fuel, contributes 19.1%. 3.2.2. PM2.5 With regards to the PM2.5 fraction, the same statistical analysis procedure was used as was applied to the PM1 fraction. The KMO and MSA criteria were evaluated in order to determine the appropriateness of FA, whereas a suitable number of factors was determined using PCA, factors with appertaining variables were defined using FA, and finally the contributions from all the factors were enumerated with regard to a given PM fraction. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.86, reflecting a very good correlation among the variables; MSA > 0.6, indicating that the ith variable is at least moderately predicted by the others). As is seen in Fig. 8, we are able to select the number of principal components to be 3 (Eigenvalue > 1).
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
403
Fig. 7. Contributions of factors to the creation of PM1.
Fig. 8. Graph of ‘‘foothills” eigenvalues – PCA (PM2.5).
In this case, it is also possible to utilize a biplot (Fig. 9) to specify the number of available principal components for use in the ongoing FA. As for PM1 we can also consider two principal components for PM2.5. These two principal components define 65% of the total variability in the original data. When three factors were selected, the third factor was repeatedly created by the chemical elements Al and V, whereas when they were grouped into two factors, those elements were partially added to the second factor, which was consequently more specific and easier to interpret. Thus, two factors were chosen for the purposes of the FA. The FA uncovered significant contributions from the two factors in the creation of PM2.5. Based on the spread of particular elements, these factors were interpreted as Factor 1, representing non-exhaust traffic sources and road dust (Fig. 10), and Factor 2, representing diesel fuel (Fig. 11) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). Particulate matter PM2.5 is most closely associated with Factor 1, representing non-exhaust traffic sources and road dust. PM2.5 most nearly correlates with chemical elements (metals) which are associated with Factor 1. The estimated contribution from each factor towards the creation of PM2.5 are depicted in Fig. 12. Factor 1, representing non-exhaust traffic sources and road dust, contributes 83% to the creation of PM2.5, and Factor 2, representing diesel fuel, contributes 17% with respect to the explained original variability in the data of variables by the three factors amounting to the value of 65%. The PM1–2.5 fraction represents only a small part of the total PM10 fraction. PM2.5 contains the PM1 fraction and PM1–2.5 as well. The ratio of fractions PM1/PM2.5 is 0.81 on average, and thus the character of the PM2.5 fraction is largely influenced by the PM1 fraction.
404
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 9. Principal components biplot (PM2.5).
Fig. 10. Factor loadings of metals on Factor 1 – F1 (PM2.5).
3.2.3. PM10 The PM10 fraction was also subjected to multivariate statistical analysis to quantify the contribution of each factor to the formation of PM10. The statistical analyses of PCA, FA and APCS were conducted in steps, performed in conjunction with the MRA. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.88, reflecting very good correlation among the variables; MSA > 0.7, indicating the ith variable is well predicted by the others). As observed in Fig. 13, we are able to select the number of principal components to be 3 (Eigenvalue > 1). These three principal components account for 80% of the total variability in the original data. Three factors were selected for the FA that resulted from the biplot (Fig. 14). As for the FA in the fraction of PM10, three factors were identified that contribute to the creation of PM10. The former complex factor ‘‘non-exhaust traffic sources and road dust”, which represented the greatest contribution to the fractions PM1 and PM2.5, was divided into two factors by the PM10 analysis. One is interpreted as a source of PM10 from ‘‘tires and road dust” (Fig. 15), and the second one is interpreted as coming from ‘‘brakes and road surfaces” (Fig. 16) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). This segregation is caused by adding the coarse fraction of PM2.5–10 into the analysis as a part of the PM10 fraction. In
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 11. Factor loadings of metals on Factor 2 – F2 (PM2.5).
Fig. 12. Contribution of factors to the creation of PM2.5.
Fig. 13. Graph of ‘‘foothills” eigenvalues based on PCA (PM10).
405
406
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 14. Principal components biplot (PM10).
Fig. 15. Factor loadings of metals on Factor 1 – F1 (PM10).
Fig. 16. Factor loadings of metals on Factor 2 – F2 (PM10).
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
407
Fig. 17. Factor loadings of metals on Factor 3 – F3 (PM10).
Fig. 18. Contribution of factors to the creation of PM10.
the coarse fraction, some of the following chemical elements dominate (Na, Al, Fe, Mn, Mo, Ba) (Fig. 2), and their weight brought the other factor into the calculation. The third factor contributing to the source characteristics of PM10 was named ‘‘diesel fuel,” similarly to the previous two analyses involving PM1 and PM2.5 (Fig. 17). Particulate matter PM10 is most associated with Factor 1, representing tires and road dust. Factor 1 is the strongest factor and characterizes the largest fraction of the variance in the original data contained in the input matrix. The approximate contributions of the individual factors to the creation of PM10 are shown in Fig. 18. Factor 1, representing tires and road dust, contributes 52.1% to the creation of PM10; Factor 2, representing brakes and road surfaces, contributes 17%, Factor 3, representing diesel fuel, contributes 14.2%, and the unresolved fraction remained at the level of 16.7%. The contributions of the individual factors to the formation of PM10 are set with respect to the original scatter in the data explained by the three factors, which amounts to the value of 80%. 4. Conclusions The identification of sources of particulate matter in the vicinity of an urban road in Zˇilina was investigated in this work. To serve the purpose of particulate matter identification, the following chemical elements (metals) were used, Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, and Pb, which may originate from the vehicular traffic. In terms of the coarse fraction PM2.5–10, the following elements were the most common: Na, Al, Fe, Mn, Mo, and Ba. In terms of the fine fraction PM2.5, the following elements were the most common: Zn, As, Cd, and Pb. Two factors were identified for the PM1 fraction, which made the following percentage contributions to its creation and were associated with particular elements: Factor 1, representing non-exhaust traffic sources and road dust (80.9%; Na, Fe, Mn, Ni, Zn, Mo, Sb, Cd, and Pb) and Factor 2, representing diesel fuel (19.1%; Mg, Ca, and Ba). The same factors were identified for PM2.5 fraction as for the PM1 fraction with the same significant elements: Factor 1, representing non-exhaust traffic sources and road dust, contributes 83% to the creation of PM2.5. Factor 2, representing diesel fuel contributes 17%.
408
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
The identification of sources based on the presence of metals in the PM10 fraction, which made the following percentage contributions to its creation and were associated with particular elements: Factor 1, representing tires and road dust (52.1%; Na, Cr, Ni, Zn, Cd, and Pb), Factor 2, representing brakes and road surfaces (17%; Al, Fe, Cu, Mo, Sb, and Ba) and Factor 3, representing diesel fuel (14.2%; Mg and Ca). The contributions of particular factors to the creation of particulate matter of the fractions PM1, PM2.5 and PM10 are closely linked to the explained part of the variability in the original variables and are determined based on the representation of selected metals in certain fractions. The fractions PM1 and PM2.5 are most associated with Factor 1, representing non-exhaust traffic sources and road dust. This factor is characterized by the chemical elements (metals) Na, Ni, Zn, Cd, Pb, Cd, Fe, Mn, Mo, and Sb. Fraction PM10 is most associated with Factor 1, representing tires and road dust, which is characterized by the metals Na, Cr, Ni, Zn, Cd, and Pb. Non-exhaust traffic sources and road dust may be considered as the predominant sources of the particulate matter making up PM1, PM2.5 and PM10 at the monitoring station described in this study in the Zˇilina city centre. Acknowledgements The paper originated thanks to support from Scientific Grant Agency of the Ministry of Education, Science, Research and Sport of the Slovak Republic and Slovak Academy of Sciences from a grant for the scientific research project VEGA 1/0557/14. References Act of the National Council of the Slovak Republic No. 137/2010 on the Air and Atmosphere. Balachandran, S., Meena, B.R., Khillare, P.S., 2000. Particle size distribution and its elemental composition in the ambient air of Delhi. Environ. Int. 26, 49–54. Chen, X., Xia, X., Zhao, Y., Zhang, P., 2010. Heavy metals concentrations in roadside soils and correlation with urban traffic in Beijing, China. J. Hazard. Mater. 181, 640–646. Durza, O., 2003. The utilization of soil magnetometry in the environmental geochemistry of heavy metals. Acta Geol. Univ. Comenianae 58, 29–55 (in Slovak). European Environment Agency, 2013. Air Quality in Europe – 2013 Report. EEA, Copenhagen, Denmark, 112 p. Gatari, M.J., Boman, J., Wagner, A., Janhall, S., Isakson, J., 2006. Assessment of inorganic content of PM2.5 particles sampled in a rural area north east of Hanoi. Sci. Total Environ. 368, 675–685. Guo, H., Wang, T., Louie, P.K.K., 2004. Source apportionment of ambient non-methane hydrocarbons in Hong Kong: application of a principal component analysis/absolute principal component scores (PCA/APCS) receptor model. Environ. Pollut. 129, 489–498. Han, L., Zhuang, G., Cheng, S., Wang, Y., Li, J., 2007. Characteristics of re-suspended road dust and its impact on the atmospheric environment in Beijing. Atmos. Environ. 41 (35), 7485–7499. Jandacka, D., Durcanska, D., 2014. Particle pollution in urban and no-urban area. In: 14th Multidisciplinary Scientific GeoConference SGEM 2014. STEF92 Technology Ltd., Sofia, Bulgaria, pp. 467–474. Kachigan, S.K., 1991. Multivariate Statistical Analysis: A Conceptual Introduction, second ed. Kukutschova, J., Moravec, P., Tomasek, V., Matejka, V., Smolík, J., Schwarz, J., Seidlerova, J., Safarova, K., Filip, P., 2011. On airborne nano/micro-sized wear particles released from low-metallic automotive brakes. Environ. Pollut. 159, 998–1006. Liati, A., Dimopoulos Eggenschwiler, P., Müller Gubler, E., Schreiber, D., Aguirre, M., 2012. Investigation of diesel ash particulate matter: a scanning electron microscope and transmission electron microscope study. Atmos. Environ. 49, 391–402. Licbinsky, R., Frybort, A., Huzlik, J., Adamec, V., Effenberger, K., Mikuska, P., Vojtesek, M., Krumal, K., 2010. Usage of scanning electron microscopy for particulate matter sources identification. Trans. Transp. Sci. 3 (3), 137–144. Lu, X., Wang, L., Li, L.Y., Lei, K., Huang, L., Kang, D., 2010. Multivariate statistical analysis of heavy metals in street dust of Baoji, NW China. J. Hazard. Mater. 173, 744–749. Manly, B.F.J., 2004. Multivariate Statistical Methods: A Primer. CRC Press, Boca Raton, USA. Manta, D.S., Angelone, M., Bellanca, A., Neri, R., Sprovieri, M., 2002. Heavy metals in urban soils: a case study from the city of Palermo (Sicily), Italy. Sci. Total Environ. 300, 229–243. McCullum, K., Kindzierski, W., 2001. Analysis of Particulate Matter Origin in Ambient Air at High Level. University of Alberta, Edmonton, Alberta. Meloun, M., Militky, J., 2006. Compendium of Statistical Data Processing. Academia, Prague (in Czech). Meloun, M., Militky, J., Hill, M., 2012. Statistical Analysis of Multidimensional Data in the Examples. Academia, Prague (in Czech). Pant, P., Harrison, R.M., 2013. Estimation of the contribution of road traffic emissions to particulate matter concentrations from field measurements: a review. Atmos. Environ. 77, 78–97. Sanderson, P., Delgado-Saborit, J.M., Harrison, R.M., 2014. A review of chemical and physical characterisation of atmospheric metallic nanoparticles. Atmos. Environ. 94, 353–365. Song, Y., Xie, S., Zhang, Y., Zeng, L., Salmon, L.G., Zheng, M., 2006. Source apportionment of PM2.5 in Beijing using principal component analysis/absolute principal component scores and UNMIX. Sci. Total Environ. 372, 278–286. Spencer, N.H., 2013. Essentials of Multivariate Data Analysis. CRC Press, Boca Raton, USA. STN EN 12341: 2016 Ambient Air Standard Gravimetric Measurement Method for the Determination of the PM10 or PM2.5 Mass Concentration of Suspended Particulate Matter. STN EN 14902: 2006 Ambient Air Quality. Standard Method for the Measurement of Pb, Cd, As and Ni in the PM10 Fraction of Suspended Particulate Matter. Tecer, L.H., 2013. Comparison and seasonal evaluation of the rural and urban ambient PM2.5 and PM10 mass concentrations based on meteorological parameters. In: 13th Multidisciplinary Scientific GeoConference SGEM 2013. STEF92 Technology Ltd., Sofia, Bulgaria, pp. 593–600. Thorpe, A., Harrison, R.M., 2008. Sources and properties of non-exhaust particulate matter from road traffic: a review. Sci. Total Environ. 400, 270–282. Tiwari, S., Bisht, D.S., Srivastava, A.K., Pipal, A.S., Taneja, A., Srivastava, M.K., Attri, S.D., 2014. Variability in atmospheric particulates and meteorological effects on their mass concentrations over Delhi, India. Atmos. Res. 145–146, 45–56. Varmuza, K., Filzmoser, P., 2009. Introduction to Multivariate Statistical Analysis in Chemometrics. CRC Press, Boca Raton, USA. Vojtesek, M., Mikuska, P., Vecera, Z., 2009. Presence, Sources and Classification of metals present in the Air. Chem. Lett. 103, 136–144. Weinbruch, S., Ebert, M., 2004. Source Apportionment of Atmospheric Aerosols Based on Electron Microscopy. Technical University of Darmstadt, Darmstadt. Yang, Z., Lu, W., Long, Y., Bao, X., Yang, Q., 2011. Assessment of heavy metals contamination in urban topsoil from Changchun City, China. J. Geochem. Explor. 108, 27–38.
Contents lists available at ScienceDirect
Transportation Research Part D journal homepage: www.elsevier.com/locate/trd
The contribution of road traffic to particulate matter and metals in air pollution in the vicinity of an urban road Dusan Jandacka a,⇑, Daniela Durcanska a, Marek Bujdos b a b
Department of Highway Engineering, Faculty of Civil Engineering, University of Zilina, Univerzitna 8215/1, 010 26 Zilina, Slovakia Comenius University in Bratislava, Faculty of Natural Sciences, Institute of Laboratory Research on Geomaterials, Mlynska dolina, 842 15 Bratislava 4, Slovakia
a r t i c l e
i n f o
Keywords: Particulate matter (PM) Sources of PM Air pollution Metals Statistical methods
a b s t r a c t A detailed investigation was conducted to study the sources of particulate matter in the vicinity of an urban road in Zˇilina. To determine the amount of particulate matter (PM10, PM2.5 and PM1) present in the ambient air, a reference gravimetric method was used. The main objective of this contribution was to identify the sources of these particles by means of statistical methods, specifically principal component analysis (PCA), factor analysis (FA), and absolute principal component scores (APCS), as well as using the presence of 17 metals in the particulate matter (Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, Pb). To identify the metals in the particulate matter samples and to determine their abundances, spectroscopic methods were used, specifically inductively coupled plasma mass spectrometry (ICP-MS). Each of these metals may come from a specific source, such as the burning of fossil fuels in fossil fuel power plants; local heating of households; the burning of liquefied fossil fuels in the combustion engines of vehicles; the burning of coal and wood; non-combustion related emissions resulting from vehicular traffic; resuspension of traffic-related dust; and industry. Diesel vehicles and non-combustion emissions from road traffic have been identified as two key sources of the particulate matter. The results reveal that non-combustion emissions, which are associated with the elements Na, Fe, Mn, Ni, Zn, Mo, Sb, Cd, and Pb, are the major contributors, followed by combustion emissions from diesel vehicles, which are associated with the elements Mg, Ca, and Ba. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Emissions of particulate matter from different sources create a very complex mixture in the air in both qualitative and quantitative terms. Their chemical compositions (in terms of chemical elements and compounds) is the result of the distribution of all the sources in space and time and the magnitudes and characteristics of the pollutants on the one hand, and meteorological and climatic conditions on the other (Tecer, 2013; Tiwari et al., 2014; Jandacka and Durcanska, 2014; Licbinsky et al., 2010). Heavy metals are among the most basic groups of contaminants that are monitored in various parts of the environment (Balachandran et al., 2000; Chen et al., 2010). The subjects of monitoring, pursuant to the general law (Act of the National Council No. 137/2010), are the following elements: As, Cd, Hg, Pb and Ni. These are generally considered as the most harmful to people and animals. These heavy metals may prove to be highly toxic to living organisms, and at the same time they are ⇑ Corresponding author. E-mail address: [email protected] (D. Jandacka). http://dx.doi.org/10.1016/j.trd.2016.11.024 1361-9209/Ó 2016 Elsevier Ltd. All rights reserved.
398
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
very persistent in the outdoor environment (that is, they do not decompose over time through environmental processes), while at the same time, they bio-accumulate in the food chain (EEA, 2013). Some of the other elements may also be dangerous; these elements can be found in soils, where they are necessary in small amounts. However, when accumulated in large quantities, they may have a toxic impact. The following elements may have these characteristics: Cr, Co, Sn, Sb, Cu, Ni, Ag, Au, Zn, Mo, V, Mn, Fe and others (Durza, 2003). These elements (metals) are bound up with fine particles of aerosols. Heavy metals enter the environment via natural and anthropogenic processes (Weinbruch and Ebert, 2004; McCullum and Kindzierski, 2001; Gatari et al., 2006; Vojtesek et al., 2009; Thorpe and Harrison, 2008; Pant and Harrison, 2013; Sanderson et al., 2014; Kukutschova et al., 2011). Natural sources include in situ weathering processes and atmospheric deposition of metals, oceanic processes and volcanic eruptions. Anthropogenic sources include the burning of fossil fuels in order to generate electricity, raw materials excavation and ore processing, industrial processes, agricultural activities, local combustion and the continuously increasing usage of motor vehicles. In particular, vehicular traffic is the sole contributor to the presence of particulate matter in the vicinity of urban roads in medium-size cities (Chen et al., 2010). This contribution characterizes the issue of air pollution involving particulate matter in the urban environment while it attempts to identify its main sources. As the main source of pollution, we shall therefore consider vehicular traffic, which produces combustion-related emissions, non-combustion related emissions, and road dust. The profiles of metals in particulate matter was used to identify the sources in an urban area. The links between the different chemical elements (metals) were identified by using multivariate statistical methods, specifically PCA, FA, and APCS, in conjunction with Multivariate Regression Analysis (MRA) (Kachigan, 1991; Manly, 2004; Meloun and Militky, 2006; Meloun et al., 2012; Spencer, 2013; Varmuza and Filzmoser, 2009; Lu et al., 2010; Yang et al., 2011; Manta et al., 2002; Guo et al., 2004; Song et al., 2006). Based on the presence of metals in the resulting groups (factors), these groups were identified as sources of PM. 2. Method of measurements and analysis 2.1. Study area The monitoring station was situated in Zˇilina city centre and on Vojtech Spanyol Street. This street represents one of the arteries connecting the secondary and tertiary city ring roads and connects the city centre with the largest housing estate of Vlcˇince. This is reflected in the amount and frequency of traffic which passes along this city radial road. The traffic volume on certain days amounts to 15,000 vehicles/24 h. The adjacent buildings along the road include housing developments and the civil amenities and facilities. The buildings create barriers on both sides of the road. We can consider this bounded space to be a street canyon. The ventilation of the canyon is limited in two ways. There is often a large accumulation of pollution during bad weather conditions, which may include inversions and windless conditions, etc. Walkways for the pedestrians occupy both sides of the monitored road. The monitoring devices were situated close to the edge of the road. 2.2. Sample collection Sampling of particulate matter was performed near the urban connecting route during 2010 (19–25 October), 2011 (8–14 March, 11–17 April, 7–14 July, 13–19 October), and 2012 (26 January–1 February, 16–22 April, 7–13 June). The goal was long-term monitoring of a representative sample of particulate matter in the atmosphere and its behaviour relative to environmental conditions. In the second phase of the research, a chemical analysis of the particulate matter was performed, as well as the determination of its possible sources. To establish the amount of particulate matter present in the ambient air, a reference method (the gravimetric method) was applied, pursuant to the standards of STN EN 12341 (2016). The sampling was performed using low volume flow samplers (LECKEL LVS3, Low Volume Samplers). In total, 3 pieces were used. Three fractions of particulate matter were monitored concurrently, specifically PM10, PM2.5 and PM1. The coarse fraction PM2.5–10 was determined as the difference between fractions PM10 and PM2.5, and the PM1–2.5 fraction was determined as the difference between fractions PM2.5 and PM1. Particulate matter was collected on nitrocellulose filters having a diameter of 47 mm at a fixed airflow rate of 2.3 #Nm3/h (normal cubic meters per hour) under standard conditions of 101.325 kPa and 0 °C, and then the mass of particulate matter collected on the filters was determined. All filters used were conditioned at 20 °C and 48.2% RH prior to sampling, as well as after sampling and weighing (STN EN 12341, 2016). The particulate matter was sampled on the filters for 16 h during daytime sampling (6 am–10 pm) and for 8 h during night sampling (10 pm–6 am). The filters were changed in the samplers twice every 24 h (at 6 am and at 10 pm). At the end of sampling, we had 108 samples of each PM fraction eligible for further analysis. Alongside the monitoring of PM, basic meteorological data were observed as well, including temperature, ambient humidity, speed and direction of wind, and precipitation. 2.3. Sample preparation and analysis The particulate matter contains various elements and compounds. In the next phase of the research, we concentrated on the measurement of the selected metals found in the fractions PM10, PM2.5 and PM1. We focused on monitoring 17 metals
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
399
which we selected based on the assumption and probabilities that they originate in the vehicular traffic (Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, Pb). Each of these metals may come from a specific source (Weinbruch and Ebert, 2004; McCullum and Kindzierski, 2001; Gatari et al., 2006; Vojtesek et al., 2009; Thorpe and Harrison, 2008; Pant and Harrison, 2013; Sanderson et al., 2014; Kukutschova et al., 2011). To identify and to determine the amounts of chemical elements (metals) in the samples of particulate matter, spectroscopic methods were utilized (ICP-MS). The analyses of the 108 filters of each PM fraction and the determination of metals present in the fractions PM1, PM2.5 and PM10 were performed pursuant to the standard STN EN 14902 (2006). All solutions were prepared using deionized water (Type 1, Labconco WaterPro). Reagents used for sample treatment included 65% nitric acid and 40% hydrofluoric acid (all Suprapur, Merck, Germany). Calibration standards were prepared from 100 mg/l batch standard solution Astasol-MIX CZ9090 MN1 (Analytika, Czech Republic) and 1000 mg/l Sb standard solution (CertiPUR, Merck, Germany). Nitrocellulose membrane filters with the sampled airborne particulate matter were placed into 15 ml polypropylene test tubes. 2.0 ml of HNO3 and 0.67 ml of HF was added, and the tubes were sealed. Samples were left for 14 days at laboratory temperature to dissolve with one daily agitation. Subsequently, 7.33 ml of deionized water was added to each sample (the final volumes were 10 ml), and the undissolved residues were removed from the digests by centrifugation (10 min at 3000g). Working solutions were prepared by dilution of the digests and 10 ppb Rh was added as an internal standard. The measurements were performed using a quadrupole ICP-MS spectrometer (Perkin-Elmer Sciex Elan 6000). 3. Results and discussion 3.1. Particulate matter and heavy metals concentrations The overall number of measurements and the data on particulate matter concentration used represents 108 measurements for each fraction, PM1, PM2.5 and PM10. The average concentration was calculated from the day and night concentrations of PM in a given sampling period (i.e., 7 daily values and 7 night values were observed for each fraction). Maximum concentrations of PM were observed during the sampling period in January 2012 (average concentrations for 7 days PM1 = 83.8 lg/m3, PM2.5 = 112.6 lg/m3, PM10 = 125.0 lg/m3), and the minimum values were observed during the sampling period in June 2012 (average concentrations for 7 days PM1 = 12.0 lg/m3, PM2.5 = 17.2 lg/m3, PM10 = 22.4 lg/m3). The sampling period of January 2012 was accompanied by very low temperatures (the lowest daily average temperature was 11.5 °C) and frequent inversions occurred over the entire sampling period during which measurements were taken, which may to a large extent affect the PM emissions produced in terms of their accumulation in the breathing zone. On the other hand, the sampling period during June 2012 was characterized by multiple rainfall events and windy weather which resulted in the ambient air having lower PM concentrations. Moreover, during this period, conditions were very suitable for dispersion and the rainfall contributed to the natural elimination of the PM from the air by means of wet deposition. The distributions of the size fractions of PM during the various sampling periods were equally variable. The fine fraction PM2.5 had the highest share of the PM10 fraction (90%) during the sampling period in January 2012 (Fig. 1). The coarse fraction, PM2.5–10, achieved the highest fraction of PM10 during the period of April 2011, when this value reached 35% of the PM10 (Fig. 1). On average, during all the monitoring periods, the fine fraction of PM2.5 amounted to 77% by weight of PM10 (the ratio PM2.5/PM10 was 0.77, on average) (Tiwari et al., 2014). Similar results have been observed over the whole of Europe, whereas it is the fine fraction of PM2.5 which has been shown to have negative impacts on the health of the population (EEA, 2013). A number of metals are well represented in the coarse fraction (>60%), namely Na, Al, Fe, Mn, Mo, and Ba. Metals well represented in the fine fraction (>60%) include Zn, As, Cd, and Pb. The metals which are approximately equally represented in both fractions are Mg, V, Ni, Sb, Ca, Cr, and Cu (Fig. 2).
Fig. 1. Distribution of PM fractions in the total PM10 fraction during different months.
400
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 2. The average concentrations and presence of metals in the PM fractions.
3.2. Identification of particulate matter sources Multivariate statistical analyses using PCA and FA were used for statistical assessment (Kachigan, 1991; Manly, 2004; Meloun and Militky, 2006; Meloun et al., 2012; Spencer, 2013; Varmuza and Filzmoser, 2009; Lu et al., 2010; Yang et al., 2011; Manta et al., 2002; Guo et al., 2004; Song et al., 2006). The appropriateness of the use of factor analysis is proven by the KMO test (Kaiser-Meyer-Olkin) and MSA (Measure of Sampling Adequacy) (Meloun et al., 2012). To approximate the contributions from specific sources of PM, the method of APCS in conjunction with MRA (Guo et al., 2004; Song et al., 2006) is adopted. The input data for the purposes of identification of possible PM sources consisted of the concentrations of particular chemical elements, specifically metals (ng/m3), as well as the concentrations of PM (lg/m3). The concentration of particulate matter was added as a variable to the input matrix in order to detect which factor is most associated with each fraction of the particulate matter. The data matrix consisted of 18 variables and 108 items, for a total of (18 108) data objects. Multivariate statistical analyses and source identification were processed separately for the fractions PM1, PM2.5 and PM10. 3.2.1. PM1 To establish the appropriateness of the factor analysis adopted here, the KMO (interval 0–1) and MSA (interval 0–1) criteria were calculated. In accordance with these criteria, the adoption of factor analysis is substantiated. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.82, reflecting very good correlation among the variables; MSA > 0.5, indicating that the ith variable is at least moderately predicted by the others). When it is deemed to be necessary to specify the number of factors before the factor analysis calculation can be run, the analysis of principal components was performed beforehand. As a result, we could determine the number of principal components which, to a sufficient degree, account for the variability in the data. As is seen in Fig. 3, we are able to fix the number of principal components at 4 (Eigenvalue > 1). The biplot shown in Fig. 4 simplifies our decision-making process over the course of selecting the number of factors for use in factor analysis, where close relationships among several variables can be observed. In this case, there are two collections of variables separated by a significant disparity. That is, there are 2 principal components which may be consecutively used in factor analysis. Two factors were selected for factor analysis. These two principal components account for 60.5% of the total variability in the original data. The third and fourth factors were eliminated, which resulted from the consecutive factor analysis using the four factors. The fourth factor was made up primarily of one element, chromium, and that is why it represented a specific factor which is at the same time difficult to define and interpret. The third factor was composed of the elements of aluminium and vanadium, whereas following the reduction of factors; these two elements were partially moved into the second factor. In fact, following that reduction, the second factor had a more apparent form, and it became feasible to identify it. The Varimax model was used for factor rotation. The factor loadings associated with variables may be explained as the correlations between the factors and the variables. They represent the most important unit of information that the interpretation of the factors is based on. Each factor contains information about contributions of selected elements (metals). As the
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
401
Fig. 3. Graph of ‘‘foothills” eigenvalues based on PCA (PM1).
Fig. 4. Principal components biplot (PM1).
most decisive loading factors, values close to or >0.7 were selected. Based on the representation and presence of elements in particular factors, the following factors may be named (Figs. 5 and 6). The factors (pollution sources) for PM1 (Figs. 5 and 6), PM2.5 (Figs. 10 and 11) and PM10 (Figs. 15–17) were identified based on the origins of metals, the characteristics of the measuring station, the distributions of the chemical elements in the PM fractions (Fig. 2), and seasonal changes in the concentrations of metals and PM. Tyre tread, a source of airborne particles, contains natural rubber copolymers such as styrene-butadiene rubber and polyisoprene rubber, and zinc (Zn) is added to tyre tread as zinc oxide and organozinc compounds to facilitate the vulcanization process. The chemical element Zn was chosen as marker for tyre tread (Thorpe and Harrison, 2008; Pant and Harrison, 2013). A number of metals are used extensively in brake lining materials. Fe, Cu, Ba, and Sb have been commonly identified in brake
402
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 5. Factor loadings of metals on Factor 1 – F1 (PM1).
Fig. 6. Factor loadings of metals on Factor 2 – F2 (PM1).
dust samples (Thorpe and Harrison, 2008; Pant and Harrison, 2013). The chemical elements Mg and Ca were chosen as markers for exhaust sources, specifically from diesel fuel (Sanderson et al., 2014; Liati et al., 2012). Road surfaces are a site of deposition for particles from a wide variety of sources. A fraction of the PM resulting from abrasion processes will be sufficiently large to settle out under gravity and deposit on the road surface. Road dusts and roadside soils often contain significant concentrations of a range of metals, including Pb, Cu, Cd, and Zn, which are indicative of contamination by road vehicle emissions. Ions such as Ca2+, SO24 , Cl , NO3 , K+, and Na+ were also found to be present in re-suspended dust (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Han et al., 2007). Moreover, the variables (elements) with factor loadings of <0.7 do contribute to a given factor by virtue of their weight and may facilitate its naming. Factors obtained via factor analysis of PM1 were interpreted as Factor 1, representing nonexhaust traffic sources and road dust (Fig. 5), and Factor 2, representing diesel fuel (Fig. 6) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). Particulate matter PM1 is most associated with Factor 1, representing non-exhaust traffic sources and road dust. PM1 correlates most strongly with the chemical elements (metals) which characterize Factor 1. The estimated contributions of the particular factors to the creation of PM1 are given in Fig. 7 with respect to the explained original variability in the data by the three factors, amounting to the value of 60%. Factor 1, representing nonexhaust traffic sources and road dust, contributes 80.9% to the creation of PM1 and Factor 2, representing diesel fuel, contributes 19.1%. 3.2.2. PM2.5 With regards to the PM2.5 fraction, the same statistical analysis procedure was used as was applied to the PM1 fraction. The KMO and MSA criteria were evaluated in order to determine the appropriateness of FA, whereas a suitable number of factors was determined using PCA, factors with appertaining variables were defined using FA, and finally the contributions from all the factors were enumerated with regard to a given PM fraction. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.86, reflecting a very good correlation among the variables; MSA > 0.6, indicating that the ith variable is at least moderately predicted by the others). As is seen in Fig. 8, we are able to select the number of principal components to be 3 (Eigenvalue > 1).
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
403
Fig. 7. Contributions of factors to the creation of PM1.
Fig. 8. Graph of ‘‘foothills” eigenvalues – PCA (PM2.5).
In this case, it is also possible to utilize a biplot (Fig. 9) to specify the number of available principal components for use in the ongoing FA. As for PM1 we can also consider two principal components for PM2.5. These two principal components define 65% of the total variability in the original data. When three factors were selected, the third factor was repeatedly created by the chemical elements Al and V, whereas when they were grouped into two factors, those elements were partially added to the second factor, which was consequently more specific and easier to interpret. Thus, two factors were chosen for the purposes of the FA. The FA uncovered significant contributions from the two factors in the creation of PM2.5. Based on the spread of particular elements, these factors were interpreted as Factor 1, representing non-exhaust traffic sources and road dust (Fig. 10), and Factor 2, representing diesel fuel (Fig. 11) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). Particulate matter PM2.5 is most closely associated with Factor 1, representing non-exhaust traffic sources and road dust. PM2.5 most nearly correlates with chemical elements (metals) which are associated with Factor 1. The estimated contribution from each factor towards the creation of PM2.5 are depicted in Fig. 12. Factor 1, representing non-exhaust traffic sources and road dust, contributes 83% to the creation of PM2.5, and Factor 2, representing diesel fuel, contributes 17% with respect to the explained original variability in the data of variables by the three factors amounting to the value of 65%. The PM1–2.5 fraction represents only a small part of the total PM10 fraction. PM2.5 contains the PM1 fraction and PM1–2.5 as well. The ratio of fractions PM1/PM2.5 is 0.81 on average, and thus the character of the PM2.5 fraction is largely influenced by the PM1 fraction.
404
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 9. Principal components biplot (PM2.5).
Fig. 10. Factor loadings of metals on Factor 1 – F1 (PM2.5).
3.2.3. PM10 The PM10 fraction was also subjected to multivariate statistical analysis to quantify the contribution of each factor to the formation of PM10. The statistical analyses of PCA, FA and APCS were conducted in steps, performed in conjunction with the MRA. On the basis of the results of the criteria, factor analysis can be applied (KMO = 0.88, reflecting very good correlation among the variables; MSA > 0.7, indicating the ith variable is well predicted by the others). As observed in Fig. 13, we are able to select the number of principal components to be 3 (Eigenvalue > 1). These three principal components account for 80% of the total variability in the original data. Three factors were selected for the FA that resulted from the biplot (Fig. 14). As for the FA in the fraction of PM10, three factors were identified that contribute to the creation of PM10. The former complex factor ‘‘non-exhaust traffic sources and road dust”, which represented the greatest contribution to the fractions PM1 and PM2.5, was divided into two factors by the PM10 analysis. One is interpreted as a source of PM10 from ‘‘tires and road dust” (Fig. 15), and the second one is interpreted as coming from ‘‘brakes and road surfaces” (Fig. 16) (Thorpe and Harrison, 2008; Pant and Harrison, 2013; Kukutschova et al., 2011). This segregation is caused by adding the coarse fraction of PM2.5–10 into the analysis as a part of the PM10 fraction. In
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 11. Factor loadings of metals on Factor 2 – F2 (PM2.5).
Fig. 12. Contribution of factors to the creation of PM2.5.
Fig. 13. Graph of ‘‘foothills” eigenvalues based on PCA (PM10).
405
406
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
Fig. 14. Principal components biplot (PM10).
Fig. 15. Factor loadings of metals on Factor 1 – F1 (PM10).
Fig. 16. Factor loadings of metals on Factor 2 – F2 (PM10).
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
407
Fig. 17. Factor loadings of metals on Factor 3 – F3 (PM10).
Fig. 18. Contribution of factors to the creation of PM10.
the coarse fraction, some of the following chemical elements dominate (Na, Al, Fe, Mn, Mo, Ba) (Fig. 2), and their weight brought the other factor into the calculation. The third factor contributing to the source characteristics of PM10 was named ‘‘diesel fuel,” similarly to the previous two analyses involving PM1 and PM2.5 (Fig. 17). Particulate matter PM10 is most associated with Factor 1, representing tires and road dust. Factor 1 is the strongest factor and characterizes the largest fraction of the variance in the original data contained in the input matrix. The approximate contributions of the individual factors to the creation of PM10 are shown in Fig. 18. Factor 1, representing tires and road dust, contributes 52.1% to the creation of PM10; Factor 2, representing brakes and road surfaces, contributes 17%, Factor 3, representing diesel fuel, contributes 14.2%, and the unresolved fraction remained at the level of 16.7%. The contributions of the individual factors to the formation of PM10 are set with respect to the original scatter in the data explained by the three factors, which amounts to the value of 80%. 4. Conclusions The identification of sources of particulate matter in the vicinity of an urban road in Zˇilina was investigated in this work. To serve the purpose of particulate matter identification, the following chemical elements (metals) were used, Na, Mg, Al, Ca, V, Cr, Fe, Mn, Ni, Cu, Zn, As, Mo, Sb, Cd, Ba, and Pb, which may originate from the vehicular traffic. In terms of the coarse fraction PM2.5–10, the following elements were the most common: Na, Al, Fe, Mn, Mo, and Ba. In terms of the fine fraction PM2.5, the following elements were the most common: Zn, As, Cd, and Pb. Two factors were identified for the PM1 fraction, which made the following percentage contributions to its creation and were associated with particular elements: Factor 1, representing non-exhaust traffic sources and road dust (80.9%; Na, Fe, Mn, Ni, Zn, Mo, Sb, Cd, and Pb) and Factor 2, representing diesel fuel (19.1%; Mg, Ca, and Ba). The same factors were identified for PM2.5 fraction as for the PM1 fraction with the same significant elements: Factor 1, representing non-exhaust traffic sources and road dust, contributes 83% to the creation of PM2.5. Factor 2, representing diesel fuel contributes 17%.
408
D. Jandacka et al. / Transportation Research Part D 50 (2017) 397–408
The identification of sources based on the presence of metals in the PM10 fraction, which made the following percentage contributions to its creation and were associated with particular elements: Factor 1, representing tires and road dust (52.1%; Na, Cr, Ni, Zn, Cd, and Pb), Factor 2, representing brakes and road surfaces (17%; Al, Fe, Cu, Mo, Sb, and Ba) and Factor 3, representing diesel fuel (14.2%; Mg and Ca). The contributions of particular factors to the creation of particulate matter of the fractions PM1, PM2.5 and PM10 are closely linked to the explained part of the variability in the original variables and are determined based on the representation of selected metals in certain fractions. The fractions PM1 and PM2.5 are most associated with Factor 1, representing non-exhaust traffic sources and road dust. This factor is characterized by the chemical elements (metals) Na, Ni, Zn, Cd, Pb, Cd, Fe, Mn, Mo, and Sb. Fraction PM10 is most associated with Factor 1, representing tires and road dust, which is characterized by the metals Na, Cr, Ni, Zn, Cd, and Pb. Non-exhaust traffic sources and road dust may be considered as the predominant sources of the particulate matter making up PM1, PM2.5 and PM10 at the monitoring station described in this study in the Zˇilina city centre. Acknowledgements The paper originated thanks to support from Scientific Grant Agency of the Ministry of Education, Science, Research and Sport of the Slovak Republic and Slovak Academy of Sciences from a grant for the scientific research project VEGA 1/0557/14. References Act of the National Council of the Slovak Republic No. 137/2010 on the Air and Atmosphere. Balachandran, S., Meena, B.R., Khillare, P.S., 2000. Particle size distribution and its elemental composition in the ambient air of Delhi. Environ. Int. 26, 49–54. Chen, X., Xia, X., Zhao, Y., Zhang, P., 2010. Heavy metals concentrations in roadside soils and correlation with urban traffic in Beijing, China. J. Hazard. Mater. 181, 640–646. Durza, O., 2003. The utilization of soil magnetometry in the environmental geochemistry of heavy metals. Acta Geol. Univ. Comenianae 58, 29–55 (in Slovak). European Environment Agency, 2013. Air Quality in Europe – 2013 Report. EEA, Copenhagen, Denmark, 112 p. Gatari, M.J., Boman, J., Wagner, A., Janhall, S., Isakson, J., 2006. Assessment of inorganic content of PM2.5 particles sampled in a rural area north east of Hanoi. Sci. Total Environ. 368, 675–685. Guo, H., Wang, T., Louie, P.K.K., 2004. Source apportionment of ambient non-methane hydrocarbons in Hong Kong: application of a principal component analysis/absolute principal component scores (PCA/APCS) receptor model. Environ. Pollut. 129, 489–498. Han, L., Zhuang, G., Cheng, S., Wang, Y., Li, J., 2007. Characteristics of re-suspended road dust and its impact on the atmospheric environment in Beijing. Atmos. Environ. 41 (35), 7485–7499. Jandacka, D., Durcanska, D., 2014. Particle pollution in urban and no-urban area. In: 14th Multidisciplinary Scientific GeoConference SGEM 2014. STEF92 Technology Ltd., Sofia, Bulgaria, pp. 467–474. Kachigan, S.K., 1991. Multivariate Statistical Analysis: A Conceptual Introduction, second ed. Kukutschova, J., Moravec, P., Tomasek, V., Matejka, V., Smolík, J., Schwarz, J., Seidlerova, J., Safarova, K., Filip, P., 2011. On airborne nano/micro-sized wear particles released from low-metallic automotive brakes. Environ. Pollut. 159, 998–1006. Liati, A., Dimopoulos Eggenschwiler, P., Müller Gubler, E., Schreiber, D., Aguirre, M., 2012. Investigation of diesel ash particulate matter: a scanning electron microscope and transmission electron microscope study. Atmos. Environ. 49, 391–402. Licbinsky, R., Frybort, A., Huzlik, J., Adamec, V., Effenberger, K., Mikuska, P., Vojtesek, M., Krumal, K., 2010. Usage of scanning electron microscopy for particulate matter sources identification. Trans. Transp. Sci. 3 (3), 137–144. Lu, X., Wang, L., Li, L.Y., Lei, K., Huang, L., Kang, D., 2010. Multivariate statistical analysis of heavy metals in street dust of Baoji, NW China. J. Hazard. Mater. 173, 744–749. Manly, B.F.J., 2004. Multivariate Statistical Methods: A Primer. CRC Press, Boca Raton, USA. Manta, D.S., Angelone, M., Bellanca, A., Neri, R., Sprovieri, M., 2002. Heavy metals in urban soils: a case study from the city of Palermo (Sicily), Italy. Sci. Total Environ. 300, 229–243. McCullum, K., Kindzierski, W., 2001. Analysis of Particulate Matter Origin in Ambient Air at High Level. University of Alberta, Edmonton, Alberta. Meloun, M., Militky, J., 2006. Compendium of Statistical Data Processing. Academia, Prague (in Czech). Meloun, M., Militky, J., Hill, M., 2012. Statistical Analysis of Multidimensional Data in the Examples. Academia, Prague (in Czech). Pant, P., Harrison, R.M., 2013. Estimation of the contribution of road traffic emissions to particulate matter concentrations from field measurements: a review. Atmos. Environ. 77, 78–97. Sanderson, P., Delgado-Saborit, J.M., Harrison, R.M., 2014. A review of chemical and physical characterisation of atmospheric metallic nanoparticles. Atmos. Environ. 94, 353–365. Song, Y., Xie, S., Zhang, Y., Zeng, L., Salmon, L.G., Zheng, M., 2006. Source apportionment of PM2.5 in Beijing using principal component analysis/absolute principal component scores and UNMIX. Sci. Total Environ. 372, 278–286. Spencer, N.H., 2013. Essentials of Multivariate Data Analysis. CRC Press, Boca Raton, USA. STN EN 12341: 2016 Ambient Air Standard Gravimetric Measurement Method for the Determination of the PM10 or PM2.5 Mass Concentration of Suspended Particulate Matter. STN EN 14902: 2006 Ambient Air Quality. Standard Method for the Measurement of Pb, Cd, As and Ni in the PM10 Fraction of Suspended Particulate Matter. Tecer, L.H., 2013. Comparison and seasonal evaluation of the rural and urban ambient PM2.5 and PM10 mass concentrations based on meteorological parameters. In: 13th Multidisciplinary Scientific GeoConference SGEM 2013. STEF92 Technology Ltd., Sofia, Bulgaria, pp. 593–600. Thorpe, A., Harrison, R.M., 2008. Sources and properties of non-exhaust particulate matter from road traffic: a review. Sci. Total Environ. 400, 270–282. Tiwari, S., Bisht, D.S., Srivastava, A.K., Pipal, A.S., Taneja, A., Srivastava, M.K., Attri, S.D., 2014. Variability in atmospheric particulates and meteorological effects on their mass concentrations over Delhi, India. Atmos. Res. 145–146, 45–56. Varmuza, K., Filzmoser, P., 2009. Introduction to Multivariate Statistical Analysis in Chemometrics. CRC Press, Boca Raton, USA. Vojtesek, M., Mikuska, P., Vecera, Z., 2009. Presence, Sources and Classification of metals present in the Air. Chem. Lett. 103, 136–144. Weinbruch, S., Ebert, M., 2004. Source Apportionment of Atmospheric Aerosols Based on Electron Microscopy. Technical University of Darmstadt, Darmstadt. Yang, Z., Lu, W., Long, Y., Bao, X., Yang, Q., 2011. Assessment of heavy metals contamination in urban topsoil from Changchun City, China. J. Geochem. Explor. 108, 27–38.