Civil aviation impacts on local air quality: A survey inside two international airports in central Italy

Accepted Manuscript Civil aviation impacts on local air quality: A survey inside two international airports in central Italy Francesca Vichi, Massimiliano Frattoni, Andrea Imperiali, Catia Balducci, Angelo Cecinato, Mattia Perilli, Paola Romagnoli PII:

S1352-2310(16)30597-0

DOI:

10.1016/j.atmosenv.2016.08.005

Reference:

AEA 14788

To appear in:

Atmospheric Environment

Received Date: 13 January 2016 Revised Date:

29 July 2016

Accepted Date: 2 August 2016

Please cite this article as: Vichi, F., Frattoni, M., Imperiali, A., Balducci, C., Cecinato, A., Perilli, M., Romagnoli, P., Civil aviation impacts on local air quality: A survey inside two international airports in central Italy, Atmospheric Environment (2016), doi: 10.1016/j.atmosenv.2016.08.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Civil aviation impacts on local air quality: a survey inside two international

2

airports in central Italy

3 4 5 6 7

Francesca Vichia§, Massimiliano Frattonia, Andrea Imperialia, Catia Balduccia, Angelo Cecinatoa, Mattia Perillia, Paola Romagnolia a Institute of Atmospheric Pollution Research, Italian National Research Council (CNR), Rome, 00016, Italy § Corresponding author

8

Abstract

RI PT

1

The results of air quality monitoring carried out over several years (2008-2012) in two international airports near Rome (labelled as A and B) are reported and discussed. Airport A serves regular flights, airport B operates low-cost flights and, during the period investigated, had about 17% of airport A aircraft traffic load. Diffusive sampling of gaseous species (NO2, SO2, BTX and O3) was performed at several sites inside the airports. During 2012 the investigation was improved by including PM10 and polycyclic aromatic hydrocarbons (PAHs). Higher concentrations of NO2 (+18%) and lower of SO2 (-20%) were found at airport B, compared to A, over the whole period investigated. The maximum concentrations of SO2 were measured in 2011 at both airports (13.4 µg/m3 and 10.8 µg/m3 respectively for A and B), despite the decrease of aircraft traffic load recorded. Statistical analysis of PM10 data showed that there was no significant difference between the average concentrations measured at the two airports (25.7 µg/m3 and 27.4 µg/m3 for A and B respectively) and among the sites investigated. The concentration of PAHs at airport B (4.3 ng/m3) was almost twice that of airport A (2.2 ng/m3), though the respective percentages of compounds were similar. Airport B seemed to be negatively influenced by its surroundings, in particular by vehicular traffic flows of two major roads, whereas airport A was positively influenced by the proximity to the seaside. PCA data analysis showed that airport A sites are differently impacted by the LTO flight phases according to their position, whereas at airport B it was impossible to find similar relationships. Keywords: Air quality; airport; diffusive sampling; polycyclic aromatic hydrocarbons

29

1. Introduction

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Civil Aviation, as a source of air pollution, has attracted a growing interest over the years since the global air traffic, according to former estimations, was expected to increase greatly in the present and the upcoming decades (Wood et al., 2008). Despite the decreasing trend due to the economic crisis recently recorded in some European countries, from 2015 onwards the forecasts for air traffic anticipate an annual growth at around 2.5% per year (EUROCONTROL – Seven Year Forecast, 2013). By consequence, the environmental impact of airport operations is still an important issue to deal with, in air quality management. At present, the impact of low-altitude aircraft emissions on local and regional air quality is still not well-known. The chemical composition of aircraft exhausts changes according to the different thrust levels employed during landing and take-off cycle (LTO) operations performed at the airports. Carbon monoxide (CO) and non-methane volatile organic compounds (NMVOC) are predominantly emitted at the low power settings or idle phases mainly occurring during the LTO cycle, so that this latter accounts for about 60% of total emissions (Tarrason et al., 2004). By contrast, approximately 95% of the total emissions of nitrogen oxides (NOx) are released during non-LTO flight phases (i.e. above 915 m and at cruise level) (Tarrason et al., 2004).

AC C

EP

TE D

M AN U

SC

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

1

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

Though total emissions of CO and NMVOC from aviation are smaller when compared to NOx, however they can play an important role in the formation of ozone surface levels, especially where the process is VOC-controlled due to low NOx levels. Non-LTO emissions of NOx affect significantly the regional air quality, however the influence of NOx emissions from LTO cycle on background concentration values near airports is not negligible. The impact of NOx emissions and subsequent photochemistry (including the formation of ozone and secondary aerosol) on public health and environment is well documented (Herndon et al., 2004; Pison et al., 2004; Carslaw et al., 2006; Wood et al., 2008). Several studies report data on direct hydrocarbons emission from aircrafts as well as average concentrations in the proximity of airports (Tesseraux, 2004; Herndon et al., 2006; Schürman et al., 2007; Jung et al., 2011). The speciation of the hydrocarbons emitted by aircraft turbine-engine shows a profile where the principal compounds are ethylene, formaldehyde, acetaldehyde, olefins and C10-paraffins (Wilkerson et al., 2010 and references therein), Despite that, no typical component or tracer of jet engine exhaust has been definitely identified. Both composition and concentration of hydrocarbon blends in aircraft exhausts look similar to those of common diesel engine exhausts (Tesseraux, 2004). A peculiarity in the composition of the exhausts may be the presence of a handful of alkanes (e.g. nonane), which are scarce in gasoline engine exhaust and have been adopted as jet engine emission marker for workers exposed to both emission types (Zeiger and Smith, 1998). Recent studies conducted inside or near airports have addressed the topic of PM emission from aircrafts (Westerdahl et al., 2008; Mazaheri et al.; 2009; Lobo et al., 2012), sometimes including organic compounds such as polycyclic aromatic hydrocarbons (PAHs) (Iavicoli et al, 2007; Hu et al., 2009; Lai et al., 2013). Concern about the fraction of ultrafine particles (UFP) emitted was also raised. The emissions of the sole aircraft source can be estimated by means of dedicated inventories using International Civil Aviation Organization (ICAO) emission factors issued for different pollutants. CO, HC, NOx and smoke number emission indices for different engines are reported in the Aircraft Engine Emissions Databank (http://easa.europa.eu/environment/edb/aircraft-engine-emissions.php). These indices are widely used in emission models for airports, but little work has been done so far to test them under real in-use conditions. Popp et al., 1999 and Herndon et al., 2004 estimated NO and NOx emission indices for different thrust levels, but compared only a few engines with ICAO data. Schäfer et al., 2003 did this comparison systematically for idle thrust and found differences between measured and certified values. The NOx emission indices were usually about 50% lower than those provided by ICAO, whilst CO emission indices were slightly higher, because thrust settings were usually lower than those defined by ICAO for idling. Mazaheri et al., 2009 have also observed differences in thrust levels during idle and taxi modes, which are considered equivalent by ICAO (7% of total thrust). The limits and uncertainties in the estimation of airport-related emissions and the growing concern caused by increasing aircraft traffic can be overcome by direct monitoring in/or nearby airports. In the present paper the results of in-field campaigns carried out at two international airports over several years are reported and discussed. The survey was performed by applying a network of diffusive samplers specific for gaseous species (NO2, SO2, BTX, i.e. benzene, toluene and xylenes, and O3). During 2012 PM10 and polycyclic aromatic hydrocarbons (PAHs) were also measured. Basic meteorological parameters and a proxy (i.e. natural radioactivity) to assess the atmospheric mixing properties provided a full picture of the periods under investigation.

AC C

45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

2

ACCEPTED MANUSCRIPT 90 91

2. Methodology 2.1 Outline of the study The two airports present different structures. Airport A, located by the seaside, covers a large area hosting three runways, whereas Airport B has only one runway, lies inland (ca. 20 km from the seaside) and is surrounded by busy roads and densely populated districts. Figures 1-2 show the layout and details of the sampling points positions at both airports.

97 98

Figure 1. Monitoring network implemented at “L. da Vinci” (airport A) (red arrow: landings, blue arrow: take offs).

RI PT

92 93 94 95 96

Figure 2. Monitoring network implemented at “G.B. Pastine” (airport B).

101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126

Runways 1 and 3 at airport A are parallel and mainly used for landings (> 97% of the total in 20082009), while runway 2 for take-offs (> 90% of the total, usually towards the sea). The single runway at airport B is positioned approximately in the same direction of the parallel runways of Airport A. The monitoring activity at the airports was designed step-by-step. In 2008 and 2009, the preliminary assessment of average concentration levels was carried out at a number of sites for gaseous pollutants regulated by the current environmental legislation. A more detailed survey was performed in 2010, including several new sites. In this case, the regular grid of sampling points was replaced by an alternative scheme. To assess the direct impact of operations carried out in the airports the sampling devices were deployed along runways, at roughly regular distances, nearby aprons and in proximity of infrastructures such as surface water runoff de-oilers, water suppliers and facilities mainly placed at the airside; further sites were chosen as indicative of the area background. During 2011 and 2012, the monitoring was continued only at some sites, selected on the basis of the previous results. The measurements were carried out in different periods each year (details in Table I – II of Supplementary Material) to account for seasonal variations. In 2012 the measurements were improved through including the PM10 fraction of airborne particulate, at sites provided with power supply. At airport A these sites were selected at increasing distance from the seaside, along SW-NE direction; they were PM1 (background, very close to the seaside), PM2 (nearby runway 2, mainly used for take offs) and PM3 (close to a taxiing area for aircrafts waiting for take-off). At airport B the sites were chosen at the northern (CD) and southern (CX) borders of the airport, near the two runway extremes. The meteorological data collected were plotted using Grapher 4.00 program (Golden Software), the statistical treatment of chemical data and multivariate analysis were performed by means of Origin Pro 8 (Origin Lab Corp) and Multibase 2014 Excel additional component, respectively.

127 128 129 130 131 132 133

AC C

EP

TE D

M AN U

SC

99 100

2.2 Air sampling and laboratory analysis Diffusive samplers (Analyst, Marbaglass, Rome, Italy) specific for different gaseous pollutants (NO2, SO2, O3 and benzene) were deployed at the selected sites. The samplers were exposed to air under aluminum shelters to protect them from direct rain, over periods varying between 15 to 30 days. The working principle and the features of the samplers are described elsewhere (De Santis et al., 1997; De Santis et al., 2002; Bertoni et al., 2001). The accuracy of diffusive sampling technique in 3

ACCEPTED MANUSCRIPT

RI PT

comparison to active, expressed as percent relative error, was found to be better than ±20% (De Santis et al., 2002). A QA/QC procedure was implemented, which implies the exposition of duplicates (about 15% of the total number of samplers) and field blanks (about 10%). The field blanks were kept closed by their cap, next to the samplers, where diffusion of the target species occurred through a stainless steel net. Blanks and duplicates were randomly positioned in order to evaluate the possible contamination during the deployment. The PM10 sampling was performed by using sequential Skypost PM instruments (Tecora, Milan, Italy), operating at a flowrate of 2.3 m3/h, each equipped with a EN LVS PM10 sampling head according to UNI EN 12341:2001 standard requirements. After gravimetric analysis of PM10, the filters were processed to determine PAHs. For this purpose, the samples were gathered into sevendays pools to reach the analytical conditions required to reliably detect PAHs according to a dedicated procedure (Di Filippo et al., 2007) even at the minimum pollution degree observed in summer. Therefore, the PM10 behavior was studied daily, as required by current legislation, while PAHs were analyzed on a weekly basis. 2.3 Analytical procedures

150

Diffusive samplers analysis

167 168 169 170 171 172 173 174 175 176 177 178 179 180

EP

TE D

The diffusive samplers were retrieved after sampling and analyzed by means of procedures already reported (De Santis et al., 1997; De Santis et al., 2002; Bertoni et al., 2001). The Analyst specific for compounds such as NO2 and O3 were extracted by adding 5 ml of 2.7 mM Na2CO3 and 0.3 mM NaHCO3 buffer directly in the sampler and stirring it for 30 minutes using VIBROMIX (203 EVT, Tehtnica) adapted to this purpose, then the solution was analyzed by means of ion chromatography (I.C.) (Dionex ICS 1000 equipped with AS12A column). For the analysis of SO2 samplers, 0.03% of H2O2 was added to the extracting solution to achieve the complete oxidation of sulfur products to sulfate. The concentrations of analytes, such as nitrate, nitrite and sulfate, were determined referring to the calibration curve set up by using solutions prepared through dilution of stock standards in water (Certipur from Merck) containing 1000 mg/L of each analyte. The BTX samplers were extracted adding in the vial 2 ml of carbon disulfide spiked with 1.0 ppmV of chlorobenzene adopted as internal standard; after 1 h, the solution was analyzed by gas chromatography coupled with flame ionisazion detection, using a J&W DB-wax column (L = 60 m, i.d. = 0.32 mm, film = 1.2 µm) provided by CPS, Milan, Italy. PAHs extraction procedure

AC C

151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166

M AN U

149

SC

134 135 136 137 138 139 140 141 142 143 144 145 146 147 148

The PM10 amounts collected on Teflon filters were determined gravimetrically by means of an electronic analytical balance (0.001 mg Sartorius-Micro, Sartorius, Milan, Italy) after conditioning 24 hours at constant temperature (20 ± 2°C) and humidity (50 ± 5%), before and after sampling. The procedure for PAHs was optimized in a previous study (Di Filippo et al., 2007), although the solvent extraction was recently revised (Cecinato et al., 2012). After collection, samples were stored in the dark at T ≤ 4°C. After solvent evaporation, the extracts were cleaned-up through chromatography on neutral alumina column; there, non-polar components were washed with isooctane, then PAHs were eluted using isooctane/dichloromethane (60:40). Solvent was changed into toluene and analytes were determined through gas chromatography coupled with mass spectrometric detection(GC-MSD) (Trace-GC Ultra coupled with DSQ-II, Thermo, Rodano, Italy). The compounds were separated by means of a EU-PAH dedicated column (CPS, Milan, Italy), and identified by MSD operated in the positive selective-ion monitoring (SIM) mode. Samples were analyzed in triplicate and the peak areas 4

ACCEPTED MANUSCRIPT

192

normalized to corresponding reference compounds (perdeuterated congeners were spiked onto samples just before solvent extraction). A certified EPA-610 PAHs mixture in toluene (Sigma, Milan, Italy) was used to prepare standard and calibration stock solutions. The relative standard deviations of the replicates ranged from 4% to 10% for all analytes and were satisfactory for our goals. The calibration curve (0.005÷4.0 µg/mL) encompassed five concentration levels of the target compounds. They fitted linear, with Pearson coefficients R2 better than 0.98. A certified material (i.e., Urban Dust SRM 1649A provided by NIST: https://wwws.nist.gov/srmors/view_detail.cfm?srm=1649A) was used to check the recovery. The method sensitivity, expressed in terms of the limit of quantification (LOQ), was better than 0.060 ng/sample for benz[a]anthracene (BaA) and 0.427 ng/sample for indeno[1,2,3-cd]pyrene (IP). No interferences were found for the target PAHs in blank filters, nor in field blanks.

RI PT

181 182 183 184 185 186 187 188 189 190 191

2.4 Meteorological parameters

The analysis of wind roses, obtained from the data of "LIRF", a meteorological station (Weather Underground Network) inside airport A, covering the entire monitoring period (2008-2012), shows the expected seasonal trend. The average wind roses, reported in Figure 3, are associated to the typical circulation of air masses over coastal areas with the alternate regime of sea-land wind provenience. Figure 3. Seasonal wind roses averaged over the entire period at Airport A.

199 200 201 202 203 204 205 206 207 208 209

The prevalence of westerly winds, i.e. towards the land, was observed during the summer. Conversely, easterly winds were prevailing during the cold season. Marked seasonal differences in wind regime were also recorded at “LIRA” (Figure 4), a meteorological station (Weather Underground Network) located at airport B. During the summer south-westerly winds were observed, whereas during the cold season the prevailing winds blew from north-east and from south/south-east. The position of the two sites with respect to the coast is different, therefore the sea-breeze blows westerly at site A and south-westerly at site B, hence the typical seasonal wind regime was observed at both sites. Both the two areas were influenced by the presence of the sea, but “LIRF” was markedly affected by the seasonal wind regime since it is located by the seaside, whereas at “LIRA” this influence was weaker due to the distance from the coast. Figure 4. Seasonal wind roses averaged over the entire period at Airport B.

210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226

In terms of intensity, winds were weaker at "LIRA" with respect to “LIRF”: about 30% and 14% of the wind records, in both warm and cold seasons, showed speeds higher than 5 m/s at airport A and B, respectively. This percentage raised up to ~ 40% in autumn 2008, and to ~ 38% in autumn and summer 2010 at airport A. Through the statistical analysis of the significance of the difference among the means of the data collected during the different seasons for each pollutant over the entire period (2009-2012) and those on a yearly basis, no relevant differences were found for both airports (see Table I, II in the Supplementary material), as it will be discussed later. Data concerning the atmospheric mixing properties were collected by measuring in-situ the natural radioactivity. Its use, as a proxy to estimate the height of the boundary layer, was previously reported (Perrino et al., 2001) and proved to be a complementary tool to evaluate the influence of meteorological factors on the concentration levels of pollutants. Since the 222Rn progeny is released from the soil with an approximately constant rate, its concentration can be considered a way to reliably estimate the dilution properties of the lower boundary layer. 222Rn concentration was determined by means of the instrument SM200 (Opsis AB, Sweden). The core of this instrument is a Geiger counter which measures the decays of the radon progeny. The counting was performed hourly on the dust collected on quartz fiber filters.

AC C

EP

TE D

M AN U

SC

193 194 195 196 197 198

5

ACCEPTED MANUSCRIPT Unfortunately, radioactivity could not be measured during 2008 and 2009. Data collected in 20102012 were compared whenever possible (Figure 5), taking into account that the measurement periods did not overlap perfectly. Figure 5. Natural radioactivity measurements performed at Airport A.

231 232 233 234 235 236 237 238 239 240 241

As expected, nocturnal maxima due to atmospheric stability and diurnal minima triggered by the dissipation of the inversion layer due to convective mixing of the lower atmosphere could be observed in the summer, whilst during the cold season a clear trend could not be recognized. In the cold months minimum values of natural radioactivity were sometimes measured during advection episodes, but prolonged stability during the daytime hours was observed as well. The summer trend was well defined in all years during July, whereas it was recognizable only during the second week (06-13 June) in 2010 and the third week (21-25 June) in 2012. Distinct comparisons for December were attempted between 2010 and 2011 as well as 2011 and 2012. It can be noticed that in 2011 hot spots of natural radioactivity were observed during the early hours of the morning, in 2010 the poor daily coverage showed very low counting values, whereas an intermediate number of counts was typical of 2012.

242

3. Results and Discussion

243 244 245 246 247 248 249

The differences between the two sites are well shown by the boxplot graphs (Figure 6), where data averaged over the entire monitoring period are represented. Descriptive statistics was also performed on the dataset (Table 1). Figure 6. Boxplot graphs of gaseous data collected over the entire period at the two airports.

SC

M AN U

TE D

Table 1. Statistical distribution of gaseous pollutants data averaged over the five years period (data are expressed in µg/m3).

EP

The significance of the difference between the two datasets, containing the concentrations averaged over the entire period at the two sites, was assessed through the use of statistical analysis (t-test, 95% confidence level). To verify whether the data met the criteria to apply parametric tests, a normality test was performed. The distribution of concentration values was normal if sites Q and T at airport A were excluded. These sites, chosen as hot spots of heavy vehicular traffic, were finally proved to be affected by very local emissions, since they were located in proximity of a parking area for diesel vehicles transporting passengers to airplanes (site Q) and near a temporary stop of tank trucks employed to supply airplanes with water (site T). Thus, they were eventually discarded from the dataset. Efforts are foreseen in future monitoring activities to correctly position the sampling points inside the apron areas. Applying the t-test, some differences in the two datasets can be expressed in quantitative terms: a statistical significance is evidenced in the differences of the means of NO2 (t(29) = -2.208, p = .035) and SO2 (t(29) = 2.271, p = .031) calculated for the two airports datasets averaged over the entire monitoring period, whereas no significant difference exists for O3 (t(29) = 1.674, p = .105) and benzene (t(29) = 0.971, p = .34). At airport A the concentration values of SO2 were higher (9.5 µg/m3) than those measured at airport B (7.9 µg/m3), whereas values of NO2 were lower (36.7 µg/m3 vs. 42.3 µg/m3). The PAHs concentrations were also compared; given the small size of the sample, the data didn’t fulfil the normality required for the application of t-test, so that the non-parametric Wilkoxon test was applied. At the 0.05 confidence level there is a statistically significant difference between the two datasets. The concentration of the PAHs averaged on a yearly basis summed up to 2.2 ng/m3 at airport A and to 4.3 ng/m3 at airport B, hence airport B had almost twice the concentration of PAHs found at airport A.

AC C

250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273

RI PT

227 228 229 230

6

ACCEPTED MANUSCRIPT Some observations concerning the surrounding area can probably explain these differences: while airport A is located in an area with a buffer zone of separation from busy roads, airport B is surrounded by two major roads along southern and western borders and is located in close proximity to a town of about 40000 inhabitants. The different type and distribution of surrounding anthropogenic sources and the mitigation due to the sea, occurring at airport A, are factors which can probably explain these findings. Yearly and seasonal trends A clear trend characterizing the period under investigation is not recognizable for the pollutants monitored. Nevertheless, comparing the yearly average concentration values (reported in Figure 7), it can be observed that the maximum values of SO2 concentrations were recorded in 2011 at both airports. Figure 7. Average yearly values calculated for different pollutants at the two airports.

287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304

To assess the difference in the means over the full dataset, the statistic test ANOVA was applied (Gratani et al., 2005). First, ANOVA provided evidence of a significant difference among the means of data grouped per year. Significantly different means were calculated for SO2 (F(4,52) = 8.69, p = 1.97 10-5) and benzene (F(4, 52) = 3.214, p = .02) and for all the pollutants measured (NO2: F(3, 29) = 3.043, p = .045; SO2: F(3, 29) = 6.65, p = .001; Benzene: F(3, 29) = 5.49, p = .004), with the exception of O3 (F(3, 29) = 2.768, p = .06), respectively for airport A and airport B datasets. Multiple comparisons can be performed using the post-hoc Tukey test (95% confidence level) (Gratani et al., 2005). It was applied to specifically assess the significance of the difference of each mean with all other means, calculated for the groups of data collected on a yearly basis (Table III and IV - Supplementary Material). By the intra-groups comparison of the concentration values performed by this test, statistically significant differences could be confirmed for SO2 at both the airports, between the year 2011 and each of the other years investigated. According to the number of aircraft movements (Table 2), the airports emissions likely decreased during 2011 at airport A and were almost constant at airport B, so that the increased SO2 concentrations recorded in 2011 cannot be ascribed to sources related to the airport activity (i.e. aircrafts, GSE and GPU). Table 2. Number of aircraft movements at the airports over the entire period.

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322

As shown by the larger number of natural radioactivity counts (Figure 5), a poor mixing of the atmosphere occurred during December 2011. Since the atmospheric mixing processes normally influence wide areas, probably both airports were affected by similar unfavorable conditions and therefore the concentrations of primary pollutants, such as SO2 and benzene, were higher in 2011 than during the other years. Since in each year of study different seasons were monitored (Table I – Supplementary Material), the corresponding datasets could be compared to identify trends and find out recurrences. At this regard, both one way ANOVA and the Tukey test were used again (Table V and VI-Supplementary Material). No relevant differences were recorded during the summer seasons of different years, apart those found for SO2 which had significantly different means at both airports over the periods considered (for airport A: F(4, 55) = 8.034, p = 3.56 10-5; for airport B: F(3, 27) = 8.823, p = 3.06 10-4). Applying the Tukey test (95% confidence level) it was found that SO2 concentration behaved as the annual trend, peaking in 2011. From the comparison of the autumn-winter periods, important differences could be observed for concentrations of O3 and benzene at both airports; ANOVA results for airport A: O3 F(4, 55) = 4.725, p = .002; Benzene F(4, 55) = 4.902, p = .002; ANOVA results for airport B: O3 F(3, 28) =

AC C

EP

TE D

M AN U

SC

RI PT

274 275 276 277 278 279 280 281 282 283 284 285 286

7

ACCEPTED MANUSCRIPT

366 367 368 369 370 371

M AN U

SC

RI PT

With the goal of identifying patterns of relation among the sites investigated, PCA analysis was carried out on the data averaged over the entire sampling period at both airports. In addition to the data collected through diffusive sampling, NOx concentrations were provided by active measurements performed by a mobile lab operating inside the hubs. The eigenvalues and loadings of the principal components (PCs) calculated through PCA analysis are reported in Table 3. The principal components, according to Kaiser criterion, are those having an eigenvalue greater than 1 (i.e. component 1 and 2 for both airports). The first component obtained explains about 50% of the variance for both the airports, but its correlation with the original variables is different for the two datasets. Component 1 is positively correlated only to ozone in the case of airport A and shows relevant anti-correlation (>0.6) vs. NOx, NO2 and benzene, whereas in the case of airport B the first component is equally positively related to these latter pollutants. The second component is in both cases negatively correlated with the original variables (especially to O3 and SO2 and for airport A also to benzene). Table 3. Eigenvalues, Contribution and Loadings of factors.

TE D

To assess the impact of aircraft operations it is important to combine the information concerning airport activities with environmental data collected through sampling. Details concerning flight operations of take-off and landing at the airports are reported in Table VII -Supplementary Material. On the basis of aircraft traffic data, as indicated by arrows in Figure 1, at airport A landings take place mainly from the north side of both runways 1 and 3, most frequently on runway 3, whereas take-offs are operated on runway 2 towards the sea. Once the components are defined by data analysis through PCA, it is possible to apply hand-drawn ellipses in which data, classified according to certain criteria (in this case the position of the sites where data were collected relative to airport operations), are comprised. Multibase additive Excel component performs this representation, though only in qualitative form. Anyway, as a matter of fact, neither protocol nor metric exists to provide a means of reporting the degree or significance of cluster obtained from PCA scores plots (Worley et al., 2013). In this case the sites were divided into five categories (as reported in Table 4): BGD (representing background levels of pollutants in the airport - green), AVIO (not representative of a flight phase, but likely mainly impacted by aircrafts emissions - blue), landing (impacted by aircraft landings lilac), take-off (impacted by aircraft take offs - red) and AUTO (likely affected by vehicular emissions - yellow). The results of this analysis for airport A are reported in Figure 8 where sites belonging to three separate groups can be observed, i.e.: i) D, M, L (lilac color); ii) S, U, R (yellow color); and iii) P, W, I (red color). Figure 8. Principal component analysis of data collected at Airport A in the whole period.

EP

344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365

3.714, p = .023; Benzene F(3, 28) = 4.636, p = .009; and also of SO2 at airport A (SO2 F(4, 55) = 12.107, p = 3.98 10-7). Also in this case, according to the Tukey test, the cold season reproduced the trend found for the different years, confirming that through the intra-groups statistical significance assessment the main differences were observed between 2011 and the other years. Areal distribution

AC C

323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343

Lilac group sites show the highest score in component 1 (bold in Table 4) which explains most of the variance (47%), the pollutants positively contributing to this factor are ozone and benzene, (see Table 4). Site D is located at the upper side of runway 1 and sites M, L in the same position on runway 3, where aircraft landings are most frequently in the initial phase. Table 4. Components scores for the different sites. 8

ACCEPTED MANUSCRIPT

RI PT

SC

M AN U

PM and PB-PAHs

TE D

The statistical analysis of PM10 data results are reported in Table 5, where, as a reference, the data of selected stations of the ARPA Lazio (Regional Agency for Environmental Protection of Lazio) network are also shown. The monitoring stations were chosen from the ARPA network for their proximity to the sites of interest, whenever possible (i.e. Ciampino), or for being representative of different surroundings (traffic, urban background, etc.) to provide information about concentration levels outside the two hubs. In Figure 9 the data distribution are reported in the form of boxplot, showing that there was no noticeable difference between the two airports and among the sites investigated.

EP

392 393 394 395 396 397 398 399 400 401 402 403 404 405 406

The red group sites, placed at the eastern side of runway 2, where aircrafts start their route for taking-off and, particularly at site P nearest to the border, are taxiing, show similar negative values of component 1 scores. This component has a significant negative correlation vs. both NOx and NO2, which are mostly emitted during take-off and climb (Masiol et al, 2014) performed in proximity to these sites. The yellow group sites are in the hub area, most affected by vehicular traffic, where support and services to flight are provided. This group is quite broad and the three sites are characterized by different values of the scores for each component. The other two ellipsoids, representing groups BGD and AVIO, are partially overlapped. The component scores for the sites of these groups are quite variable and no well-defined pattern is evident. At airport B it was impossible to define different zones according to the flight phase of aircrafts; concurrently a clear grouping of sites was not observed through PCA analysis (Figure I – Supplementary Material) In Table 3 it can be observed that component 1, explaining most of the variance (50%), is mainly influenced by NOx, NO2 and benzene, to which it is positively related. The sites with the highest score for component 1 are CE, CI and CL (bold in Table 4) located at the eastern (site CE) and at the western (sites CI and CL) side of the runway. All these sites are in proximity of external roads, for local vehicular traffic (at the eastern site), and of major roads (at the western side).

Table 5. Statistical distribution of PM10 data collected at airports in 2012 and at ARPA fixed station (data are expressed in µg/m3).

AC C

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391

407 408

Figure 9. Boxplot graphs of PM10 data collected at the airports in 2012.

409 410 411 412 413 414 415 416 417 418

According to the Kruskal-Wallis ANOVA test (95% confidence level), and taking into account that data collected at CD site were not normally distributed, PM10 did not show statistically significant difference among the sites. The analysis of the PAH content was also performed. While the sum of carcinogenic PAHs at airport B was about twofold that found at airport A (Table 6), the percent composition of the group (Figure 10) was similar for the two airports. The concentration of the regulated B(a)pyrene didn’t exceed the guideline value of European normative (i.e., 1.0 ng/m3 as yearly average) at both airports. Table 6. Yearly averages of particle bound (PB-) PAHs (data are expressed in ng/m3).

419 9

ACCEPTED MANUSCRIPT

449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467

RI PT

SC

Omitting the detailed analysis of PAH diagnostic ratios, commonly used to identify the pollution sources, attention was paid to benzo[b]fluoranthene and benzo[j]fluoranthene, whose ratio (BbF/BjF) was reported as indicative of aircrafts emissions (Tesseraux et al., 2004; Chia-Hsiang Lai et al, 2013), in the case of the predominance of the former compound. When the comparison of the two isomers (not always analytically resolved) could be made, the ratio rates suggested that this source is prevalent in both areas. Nevertheless, the BbF/BjF ratio, calculated at urban ARPA Lazio stations in the same geographic area, provided analogous results (see Table 8). Thus, the presence of PAHs in the air of airports could not be associated unequivocally to aircraft activity.

M AN U

440 441 442 443 444 445 446 447 448

Table 7. Statistical distribution of data collected at PM monitoring sites in 2012.

Table 8. B[b]F/B[j]F ratios calculated for the different sites.

The concurrent collection of many environmental parameters (including gaseous pollutants, PM10 and PB-PAHs) was performed to better characterize some sites, identify similarities among them, and confirm their preliminary classification made on the basis of the areal characteristics. In particular, multivariate analysis (PCA) was applied to the datasets collected at the three sites at airport A and at the two sites at airport B where, besides gaseous compounds, PM10 and PB-PAHs were measured. The PCA results are reported in Table 9 and plotted in Figure II – Supplementary Material. Table 9. Eigenvalues, Contribution and Loadings of factors.

TE D

428 429 430 431 432 433 434 435 436 437 438 439

The statistical treatment of the whole dataset collected at the PM10 monitoring sites is reported in Table 7. Higher concentrations of organic compounds, such as toluene, xylenes and PB-PAHs as well as PM10 and NO2, were found in 2012 at airport B, compared to airport A.

EP

422 423 424 425 426 427

Figure 10. PB-PAHs yearly averages relative abundance in PM10 collected in 2012.

The principal components, according to Kaiser criterion, are those having an eigenvalue greater than 1. Hence in this case three principal components are considered relevant. It can be observed that in the component 1 explaining most of the variance (76%), an equal load was given to all of the PB-PAHs, PM10, toluene, NOx and NO2 (Table 9). Component 2 (12% variance explained) and Component 3 (9% variance explained) have similar absolute values of SO2 loadings, but they show opposite correlation, and they are both positively correlated to benzene and toluene. In Table 10 the component scores for the different sites are reported. Negative values for component 1 were found for airport A sites (E, O and W), whereas airport B sites (CD and CX) show higher positive values. Hence the sites best represented by this component are CD and CX forming a distinct group in Figure II – Supplementary Material; sites E and W are comprised in another group, whereas site O is individual. Since component 1 is characterized by almost equivalent higher loadings for PM10, NOx and organics (toluene and PAHs) and by negative loadings in SO2 and O3, it can be concluded that sites CX and CD were dominated by various types of anthropogenic sources, including vehicular traffic, whereas sites E, W and O were mostly influenced by secondary pollutants such as ozone and likely by sources emitting SO2 (diesel vehicles).

AC C

420 421

10

ACCEPTED MANUSCRIPT Table 10. Components scores for the different sites at the two airports.

469

4. Conclusions

470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507

Important differences were observed over the years in the characteristics of the air quality of the two hubs. Higher concentrations of NO2 (+18%) and lower of SO2 (-20%) were found at airport B, compared to A, over the whole period investigated. Airport B seemed to be influenced by its surroundings, likely by vehicular traffic flows of two major roads, as well as by local traffic of the small town located in proximity of its eastern border. This finding was confirmed by higher concentrations of NO2 and PAHs found at airport B compared to Airport A, despite the lower number of aircrafts movements (about 17% of that recorded at Airport A) and passengers served. The concentration of polycyclic aromatic hydrocarbons at airport B (4.3 ng/m3) was, indeed, almost twice that of airport A (2.2 ng/m3), though the respective percentages of compounds were similar. Airport A, on the other hand, is located by the seaside and surrounded by semirural areas, hence in a suitable position in order to help dispersion and to mitigate the load of pollutants typical of an international airport with an intense aircraft traffic. Moreover it should be noticed that at Airport B, which lies close to urban centers and serves lowcost flights through a small infrastructure consisting of a single runway, it was difficult to distinguish the relative contributions of the various pollution sources. At airport A there are well defined routes for the different flight phases and it could be feasible to identify the areas affected by the same type of emissions and characterized by similar environmental conditions. Nonetheless, it is important to remember that, as highlighted by Peace et al, 2006, airports include a variety of sources, therefore these other sources should be considered along with the aircrafts. Facilities such as GSE (ground support equipment), GPU (ground power units), power plants and other treatment units are, indeed, necessary for routine operations. The identification of markers or diagnostic ratios indicative of aircraft emission is a challenging task. The use of the BbF/BjF concentration ratio as diagnostic of aircraft emissions, at least in this geographic area, didn’t prove to be conclusive. The inter-annual trends of pollutants, were not well defined nor related to the significant decrease of aircrafts movements number recorded over the five years investigated, probably due to the economic crisis. A further characterization of PM content, and an extension of the next surveys to the neighboring zones will improve the knowledge of aircraft emissions impacts on airports and their surrounding areas.

AC C

EP

TE D

M AN U

SC

RI PT

468

Acknowledgements

Thanks to Mauro Montagnoli for the help with natural radioactivity measurements, to Marco Giusto with meteorological data and to Giuliano Fontinovo for processing the maps. The authors are also grateful to Erica Perreca and Giorgio Tagliacozzo for the analysis of BTX diffusive samplers.

11

ACCEPTED MANUSCRIPT 508 509 510 511

References Bertoni, G., Tappa, R., Allegrini, I., 2001.The Internal Consistency of the 'Analyst' Diffusive Sampler - A Long-Term Field Test. Chromatographia, 54, 653–657. Carslaw, D.C., Beevers, S.D., Ropkins, K., Bell, M.C., 2006. Detecting and quantifying aircraft and

513

other on-airport contributions to ambient nitrogen oxides in the vicinity of a large international

514

airport. Atmos. Environ., 40, 5424-5434.

515 516

RI PT

512

Cecinato, A., Balducci, C., Romagnoli, P., Perilli, M., 2012. Airborne psychotropic substances in eight Italian big cities: burdens and behaviors. Environ. Pollut., 171, 140-147.

Chia-Hsiang Lai & Kuen-Yuan Chuang & Jin-Wei Chang, 2013. Characteristics of nano-/ultrafine

518

particle-bound PAHs in ambient air at an international airport. Environ. Sci. Pollut. Res. 20,

519

1772–1780.

SC

517

De Santis, F., Allegrini, I., Fazio, M.C, Pasella, D., Piredda, R., 1997. Development of a passive

521

sampling technique for the determination of nitrogen dioxide and sulphur dioxide in ambient air.

522

Anal. Chim. Acta, 34 (1), 127–134.

M AN U

520

523

De Santis, F., Dogeroglu, T., Fino, A.,· Menichelli, S., Vazzana, C., Allegrini, I., 2002. Laboratory

524

development and field evaluation of a new diffusive sampler to collect nitrogen oxides in the

525

ambient air. Anal. Bioanal. Chem., 373, 901–907.

Di Filippo, P., Riccardi, C., Gariazzo, G., Incoronato, F., Pomata, D., Spicaglia, S., Cecinato, A.,

527

2007. Air pollutants and the characterization of organic content of aerosol particles in a mixed

528

industrial/semi-rural area in central Italy. J. Environ. Monitor., 9, 275-282.

531 532

Flight Movements and Service Units 2013 – 2019 v1.0 Edition Date: 30/09/13.

EP

530

EUROCONTROL (European Organisation for the Safety of Air Navigation) Seven-Year Forecast

Gratani, L., Varone, L., 2005. Daily and seasonal variation of CO2 in the city of Rome in relationship with the traffic volume, Atmos. Environ., 39, 14, 2619-2624.

AC C

529

TE D

526

533

Herndon, S. C., Shorter, J. H., Zahniser, M. S.; Nelson, D. D., Jr., Jayne, J., Brown, R. C., Miake-

534

Lye, R. C., Waitz, I., Silva, P., Lanni, T., Demerjian, K., Kolb, C. E., 2004. NO and NO2

535

Emissions Ratios Measured from in use Commercial Aircraft during Taxi and Take-Off.

536

Environ. Sci. Technol., 38, 6078–6084.

537

Herndon, S. C., Rogers, T., Dunlea, E. J., Miake-Lye, R. C., Knighton, B., 2006. Hydrocarbon

538

emissions from in-use commercial aircraft during airport operations. Environ. Sci. Technol., 40,

539

4406–4413.

540

Iavicoli, I., Chiarotti, M., Bergamaschi, A., Marsili, R., Carelli, G., 2007. Determination of airborne

541

polycyclic aromatic hydrocarbons at an airport by gas chromatography–mass spectrometry and

542

evaluation of occupational exposure. J. Chromatogr. A, 1150, 226–235. 12

ACCEPTED MANUSCRIPT 543 544 545 546

Jung, K.-H., Artigas, F., Shin, J.Y., 2011. Personal, indoor, and outdoor exposure to VOCs in the immediate vicinity of a local airport Environ. Monit. Assess. 173, 555–567. Lai, C.H., Chuang, K.Y., Chang, J.W., 2013. Characteristics of nano-/ultrafine particle-bound PAHs in ambient air at an international airport. Environ. Sci. Pollut. Res., 20 (3), 1772–1780. Lobo, P., Hagen, D.E., Whitefield, P. D., 2012. Measurement and analysis of aircraft engine PM

548

emissions downwind of an active runway at the Oakland International Airport. Atmos. Environ.,

549

61, 114–123.

550 551

RI PT

547

Masiol, M., Harrison, R.M., 2014. Aircraft engine exhaust emissions and other airport-related contributions to ambient air pollution: A review, Atmos. Environ., 95, 409-455.

Mazaheri, M., Johnson, G.R., Morawska, L., 2009. Particle and gaseous emissions from

553

commercial aircraft at each stage of the landing and takeoff cycle. Environ. Sci. Technol. 43 (2),

554

441-446.

556

Peace, H., Maughan, J., Owen, B., Raper, D., 2006. Identifying the contribution of different airport

M AN U

555

SC

552

related sources to local urban air quality. Environ. Model. Softw., 21(4), 532–538.

557

Perrino, C., Pietrodangelo, A., Febo, A., 2001. An atmospheric stability index based on radon

558

progeny measurements for the evaluation of primary urban pollution. Atmos. Environ., 35(31),

559

5235–5244.

562 563 564 565

TE D

561

Pison, I., Menut, L., 2004. Quantification of the impact of aircraft traffic emissions on tropospheric ozone over Paris area. Atmos. Environ., 38, 971–983. Popp, P.J., Bishop, G.A., Stedman, D.H., 1999. Method for commercial aircraft nitric oxide emission measurements. Environ. Sci. Technol., 33, 1542–1544. Schäfer, K., Jahn, C., Sturm, P., Lechner, B., Bacher, M., 2003. Aircraft emission measurements by

EP

560

remote sensing methodologies at airports. Atmos. Environ. 37, 5261–5271. Schürmann, G., Schäfer, K., Jahn, C., Hoffmann, H., Bauerfeind, M., Fleuti, E., Rappenglück, B.,

567

2007. The impact of NOx, CO and VOC emissions on the air quality of Zurich airport. Atmos.

568

Environ., 41 (1), 103–118.

569 570 571 572

AC C

566

Tarrasón, L., Jonson, J. E., Berntsen, T. K., & Rypdal, K., 2004. Study on air quality impacts of non-LTO emissions from aviation. Norwegian Meteorological Institute. Tesseraux, I., 2004. Risk factors of jet fuel combustion products. Toxicol. Lett. Volume 149 (1–3), 295–300.

573

Westerdahl, D., Fruin, S., Fine, P. L., Sioutas, C., 2008. The Los Angeles international airport as a

574

source of ultrafine particles and other pollutant to nearby communities. Atmos. Environ., 42,

575

3143–3155.

13

ACCEPTED MANUSCRIPT 576

Wilkerson, J. T., M. Z. Jacobson, A. Malwitz, S. Balasubramanian, R. Wayson, G. Fleming, A. D.

577

Naiman, and S. K. Lele, 2010. Analysis of emission data from global commercial aviation: 2004

578

and 2006, Atmos. Chem. Phys., 10, 6391–6408. Wood, E. C., Herndon, S. C., Timko, M. T., Yelvington, P. E., Miake-Lye, R. C., 2008. Speciation

580

and chemical evolution of nitrogen oxides in aircraft exhaust near airports Environ. Sci.

581

Technol., 42, 1884– 1891.

583 584 585

Worley, B., Halouska, S., Powers, R., 2013. Utilities for quantifying separation in PCA/PLS-DA scores plots, Anal. Biochem., 433 (2), 102-104.

Zeiger, E. and Smith, L., 1998. The first international conference on the environmental health and safety of jet fuel. Environ. Health Persp., 106(11): 763–764.

SC

582

586

EP

TE D

M AN U

http://www.arpalazio.net/main/aria/

AC C

587

RI PT

579

14

ACCEPTED MANUSCRIPT Table 1. Statistical distribution of gaseous pollutants data averaged over the entire period (data are expressed in μg/m3).

Value

Sample

1st Quartile (Q1)

0.2

25.3

S

29.1

35.3

43.3

51.2

D

10.4

0.1

67.3

F

79.2

86.1

90.6

111.1

W

9.5

1.7

0.2

6.5

P

8.4

9.6

10.8

12.5

V

19

1.9

0.6

0.3

1.1

O

1.4

1.8

2.1

3.5

NO2

12

42.4

4.8

0.1

35.1

CG

39.0

41.7

44.7

51.5

CE

O3

12

78.9

13.9

0.2

57.0

CI

SO2

12

7.9

2.1

0.3

5.3

CC

Benze ne

12

1.7

0.3

0.2

CV

19

36.7

8.0

O3

19

86.2

SO2

19

Benze ne

Airport A Airport B

Min.

Median

1.1

3rd Quartile (Q3)

Max.

Max.

Value

Sample

RI PT

NO2

Min.

Standard Deviation

SC

Mean

V

67.8

81.9

84.4

103.5

CF

6.2

7.3

9.5

11.5

CF

1.6

1.9

2.3

CI

M AN U

N total

Site

CC

1.5

Table 2. Number of aircraft movements at the airports over the entire period.

346.435

Airport B

-

2010

2011

2012

324.497

329.269

328.496

313.850

57.585

54.040

54.714

50.666

EP

Airport A

2009

TE D

2008

AC C

Table 3. Eigenvalues, Contribution and Loadings of factors. Airport A

Airport B

Comp 1 Comp 2 Comp 3 Comp 1 Comp 2 Comp 3

Contribution

47%

28%

13%

50%

35%

9%

Accumulation of Contribution

47%

75%

89%

50%

85%

94%

Eigenvalue

2.35

1.42

0.67

2.08

1.47

0.38

NOx NO2 O3 SO2 Benzene

-0.63 -0.62 0.36 -0.27 0.13

-0.03 -0.13 -0.49 -0.59 -0.63

-0.10 0.02 0.46 0.45 -0.76

0.57 0.57 -0.23 -0.08 0.54

-0.03 -0.22 -0.66 -0.71 -0.11

-0.56 -0.19 -0.28 0.23 0.72

ACCEPTED MANUSCRIPT Table 4. Components scores for the different sites. Airport B Groups west west west east east east

Comp 1 -0.87 -1.62 -1.80 -0.10 1.68 -0.56 -1.93 -0.01 2.86 2.42 0.04 -0.11

Comp 2 -0.59 0.37 1.28 0.50 -2.52 -2.48 -0.33 -0.04 1.43 0.28 0.77 1.33

RI PT

Site CA CB CC CD CE CF CG CH CI CL CM CN

west west west east

SC

Comp 3 -0.27 0.08 -0.43 -0.47 1.16 1.05 0.41 0.24 0.07 -0.15 0.42 0.61 -2.46 0.95 0.24 0.37 -0.41 -0.57 -0.85

TE D

Site Groups Comp 2 A BGD -1.24 B BGD -0.73 C avio -0.72 D landing 0.45 E BGD 0.65 F avio -2.39 G BGD -0.29 H BGD 0.46 I take_off -1.02 L landing 0.99 M landing 1.66 N avio 0.47 O avio -1.35 P take_off -1.34 R auto 0.97 S auto 0.22 U auto 0.69 Z 2.51 W take_off 0.00 *BGD: BACKGROUND AVIO: AIRCRAFT OPERATION AREA AUTO: VEHICLES OPERATION AREA LANDING: LANDING AREA TAKE OFF: TAKE OFF AREA

Comp 3 -0.02 -0.38 -1.13 0.08 -0.08 0.01 0.71 0.12 0.60 -1.19 1.10 0.18

M AN U

Airport A Comp 1 -0.42 0.06 1.10 2.61 1.05 1.70 0.09 0.05 -1.55 1.72 1.56 0.91 0.90 -2.01 -0.12 -3.03 -1.29 -1.43 -1.93

Table 5. Statistical distribution of PM10 data collected at airports in 2012 and at ARPA fixed station (data are expressed in μg/m3).

E O W Civitavecchia Villa Ada Guido Ciampino Fermi Cipro CX CD Civitavecchia Villa Ada Guido Ciampino Fermi Cipro

64 56 55 58 61 60 62 60 61 59 42 57 58 56 58 58 58

Mean

Standard Deviation

Minimum

26.2 25.1 25.8 22.1 23.8 24.0 29.3 32.2 26.2 27.3 27.6 20.9 21.4 21.3 28.1 30.3 25.1

9.9 9.9 9.1 7.6 8.3 9.2 12.3 9.3 9.8 10.7 15.8 6.9 7.7 8.9 11.9 8.0 9.5

9.9 7.7 7.5 5.0 9.0 6.0 9.0 12.0 6.0 8.6 5.1 5.0 4.0 6.0 9.0 12.0 6.0

EP

N total

AC C

Site

1st Quartile (Q1) 19.8 17.9 20.0 17.0 18.0 16.5 20.0 26.0 19.0 19.9 17.1 16.0 16.0 15.0 20.0 26.0 19.0

Median 25.9 25.5 24.0 22.0 24.0 23.0 27.5 33.0 25.0 25.6 21.8 20.0 21.0 21.0 26.0 30.0 23.5

3rd Quartile (Q3) 32.3 29.3 30.9 26.0 28.0 32.0 35.0 37.5 31.0 33.5 35.8 24.0 26.0 24.0 35.0 36.0 31.0

Maximum

P90.4

59.8 48.5 57.0 41.0 47.0 46.0 75.0 58.0 50.0 54.5 69.9 37.0 45.0 46.0 75.0 51.0 50.0

36.5 38.6 36.9 32.0 35.0 36.0 44.0 43.0 40.0 43.6 51.1 31.0 29.0 34.0 42.0 41.0 40.0

ACCEPTED MANUSCRIPT Table 6. Yearly averages of PB-PAHs (data are expressed in ng/m3).

O 0.26 0.22 0.73 0.28

benzo(a)pyrene Perylene indeno(1,2,3-c,d)pyrene dibenzo(a,h)anthracene benzo(g,h,i)perylene

0.19 0.04 0.26 0.05 0.32

0.25 0.05 0.31 0.06 0.34

0.22 0.04 0.29 0.05 0.32

Carcinogenic PAHs

1.24

1.62

1.38

CD 0.30 0.48 1.37 0.52

CX 0.23 0.39 1.14 0.47

0.57 0.11 0.58 0.10 0.60

0.47 0.09 0.52 0.08 0.60

2.92

2.44

RI PT

E 0.08 0.17 0.65 0.25

SC

Compound benzo(a)anthracene chrysene Benzo(b+j+k)fluoranthene benzo(e)pyrene

Site W 0.10 0.21 0.72 0.28

Table 7. Statistical distribution of data collected at PM monitoring sites in 2012. Maximum

Max Sample

Minimum

Min Sample

Average

Standard Deviaton

CV

Benzene Toluene Xylenes NOx NO2 O3 SO2 PM10 benzo(a)anthracene chrysene Benzo(b+j+k)fluoranthene benzo(e)pyrene benzo(a)pyrene Perylene indeno(1,2,3-c,d)pyrene dibenzo(ah)anthracene

4.7 13.1 5.9 65.7 49.1 84.8 11.0 27.6 0.4 0.7 1.8 0.7 0.8 0.2 0.7 0.1

O CX CD CD CX E W CD CD CX CX CX CX CX CX CD

0.5 1.6 3.3 40.4 30.6 42.8 5.9 25.1 0.1 0.2 0.7 0.2 0.2 0.0 0.3 0.1

E E W O O CX O O E E E E E E E W

2.3 7.4 4.3 52.7 41.5 62.8 7.4 26.4 0.3 0.4 1.1 0.4 0.4 0.1 0.5 0.1

1.6 4.4 1.2 11.9 6.9 17.9 2.1 1.0 0.2 0.3 0.6 0.2 0.3 0.1 0.2 0.0

0.7 0.6 0.3 0.2 0.2 0.3 0.3 0.0 0.7 0.7 0.5 0.5 0.7 0.7 0.5 0.5

CX

0.3

E

0.5

0.3

0.5

AC C

EP

TE D

M AN U

Pollutant

benzo(ghi)perylene

0.9

Table 8. B(b)F/B(j)F ratios calculated for the different sites. ARPA

Airport A

Airport B

Spring

Urban Bgd 1 Cipro 2.0

Traffic Francia 2.1

Urban Bgd 2 Villa Ada 2.1

n.a.

n.a. n.a.

n.a.

n.a. n.a.

n.a.

Summer

1.8

1. 6

1.8

1.9

1.8 1.8

1.9

1.6

1.6

Autumn

-

-

-

2.1

2.0 2.0

2.0

2.2 2.1

2.2

Winter

-

-

-

1.8

1.9 1.9

1.9 1.9

1.8 1.8

1.8 1.8

Season

E

O

W

average CD CX average

ACCEPTED MANUSCRIPT Table 9. Eigenvalues, Contribution and Loadings of factors. Comp 2

Comp 3

Contribution Accumulation of Contribution

76%

12%

9%

76%

88%

97%

Eigenvalue Benzene Toluene Xylenes NOx NO2 O3 SO2 PM10 benzo(a)anthracene chrysene Benzo(b+j+k)fluoranthene benzo(e)pyrene benzo(a)pyrene Perylene indeno(1,2,3-c,d)pyrene dibenzo(ah)anthracene benzo(ghi)perylene

12.16 -0.02 0.24 0.06 0.25 0.21 -0.27 -0.12 0.25 0.25 0.28 0.28 0.28 0.28 0.28 0.28 0.28 0.28

1.96 0.63 0.03 -0.15 -0.16 -0.45 -0.05 -0.45 -0.24 0.28 0.03 0.02 0.02 0.04 0.05 0.04 0.05 0.02

TE D

M AN U

SC

RI PT

Comp 1

AC C

EP

Table 10. Components scores for the different sites at the two airports.

Site

Comp 1

Comp 2

Comp 3

E

-2.99

-1.21

-1.41

O

-2.60

2.35

0.11

W

-2.25

-1.14

1.52

CD

3.72

0.12

-1.00

CX

4.11

-0.12

0.78

1.40 0.13 0.41 -0.73 0.17 0.07 -0.21 0.40 -0.19 -0.04 -0.02 -0.03 0.00 -0.04 -0.03 -0.02 -0.08 0.00

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

Figure 1. Monitoring network implemented at “L. da Vinci” (airport A) (red arrow: landings, blue arrow: take offs).

Figure 2. Monitoring network implemented at “G.B. Pastine” (airport B).

ACCEPTED MANUSCRIPT Airport A Hot Season

Airport A Cold Season

Wind Speed (m/s) <=1 >1 - 2 >2 - 3 >3 - 4 >4 - 5 >5 - 6 >6 - 7 >7 - 8 >8

0

0

315

45

45

RI PT

315

Wind Speed (m/s) <=1 >1 - 2 >2 - 3 >3 - 4 >4 - 5 >5 - 6 >6 - 7 >7 - 8 >8

270 270

90

90 0%

4%

8%

12%

0%

16%

225 225

4%

8%

12%

16%

135

135

180

SC

180

Airport B Hot Season

TE D

Wind Speed (m/s) <=1 >1 - 2 >2 - 3 >3 - 4 >4 - 5 >5 - 6 >6 - 7 >7 - 8 >8

0

315

4%

AC C

0%

225

8%

12%

90

Airport B Cold season Wind Speed (m/s) <=1 >1 - 2 >2 - 3 >3 - 4 >4 - 5 >5 - 6 >6 - 7 >7 - 8 >8

0

315

45

EP

270

M AN U

Figure 3. Seasonal wind roses averaged over the entire period at Airport A.

45

270

90 0%

4%

8%

12%

16%

225

135

135

180

180

Figure 4. Seasonal wind roses averaged over the entire period at Airport B.

16%

ACCEPTED MANUSCRIPT 2500

2500

2010

2010

2012

2012 2000

1500

1500

2011

18/12

8/12

10/12

6/12

SC 30/11

M AN U

8/8

3/8

29/7

24/7

4/12

0

0

2/12

500

500

16/12

Counts

Counts

1000

1000

14/12

1500

1500

12/12

2012

2000

2000

19/7

26/6 2011

2012

14/7

25/6

2010

2010

9/7

24/6

RI PT

2500

2500

23/6

22/6

20/6

15/6

9/6

1/6

13/6

0

11/6

0

7/6

500

5/6

500

3/6

1000

30/5

1000

21/6

Counts

Counts

2000

Figure 5. Natural radioactivity measurements performed at Airport A. Airport A 120

14

TE D

60

40

20

0

80

8

6

4

0

O3

SO2

Benzene

Pollutant

Pollutant

Airport B

Airport B 14

12

Concentration (µg/m3)

Concentration (µg/m3)

100

NO2

AC C

NOx

10

2

EP

Concentration (µg/m3)

80

Concentration (µg/m3)

12

100

120

Airport A

60

40

10

8

6

4

2

20

0

0 NOx

NO2

Pollutant

O3

SO2

Benzene

Pollutant

Figure 6. Boxplot graphs of gaseous data collected over the entire period at the two airports.

RI PT

ACCEPTED MANUSCRIPT

M AN U

SC

Figure 7. Average yearly values calculated for different pollutants at Airport A and at Airport B.

TE D

Figure 8. Principal component analysis of data collected at Airport A in the whole period.

Airport A

60 50 40 30 20 10

AC C

Concentration (µg/m3)

70

80 70

Concentration (µg/m3)

EP

80

Airport B

60 50 40 30 20 10

0

0 E

O

W

CX

Site

Figure 9. Boxplot graphs of PM10 data collected at the airports in 2012.

CD

Site

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

Figure 10. PB-PAHs yearly average percentages in PM10 collected in 2012.

ACCEPTED MANUSCRIPT Highlights

EP

TE D

M AN U

SC

RI PT

Comparison of airports with different layout/aircraft traffic load. Higher NO2 and PB-PAHs, lower SO2 found in the smaller airport. Maximum SO2 measured at airports despite decrease of aircraft traffic recorded. BbF/BjF diagnostic of aircraft emissions didn’t prove to be conclusive.

AC C

• • • •