Water Air Soil Pollut (2011) 220:265–278 DOI 10.1007/s11270-011-0752-6
Particle Number Size Distribution and Weight Concentration of Background Urban Aerosol in a Po Valley Site Alessandro Bigi & Grazia Ghermandi
Received: 30 June 2010 / Accepted: 14 January 2011 / Published online: 20 February 2011 # Springer Science+Business Media B.V. 2011
Abstract Measurements of particle size distributions and PM2.5 from an urban background site in the Central Po Valley are analysed; the site is one of the medium–small-size cities in the central valley, without the direct influence of the metropolitan and industrial area of Milan and of the Adriatic Sea. The data comprise number concentration of particle with diameters ranging between 10 and 700 nm, PM2.5 and main meteorological variables from February to August 2008. Daily cycles of the observed pollutants are analysed, along with auto-correlation function for particle number concentration and principal component analysis (PCA) of all the available variables; finally, the diurnal pattern of PM2.5 low-, mediumand high-pollution events has been investigated. Total particle number concentration showed a daily pattern both in winter and summer, although different between weekdays and Sundays and with wider variations during the cold season. A daily cycle is present for the geometric mean diameter of nucleation mode particles in winter and of nucleation and Aitken mode particles in summer. PM2.5 showed a slight daily pattern for weekdays and Sundays, similar, but lagged, to total particle count cycle. Mixing layer depth resulted the main process controlling PM2.5, A. Bigi (*) : G. Ghermandi Department of Mechanical and Civil Engineering, University of Modena and Reggio-Emilia, via Vignolese 905/b, Modena 41125, Italy e-mail:
[email protected]
although also human activities contribute to PM2.5 concentration and allow some deposition and (re-) mobilisation at the first hours of the day and morning rush hour, respectively, while particle number concentration responds immediately to anthropogenic sources. PCA confirmed the dependence of particle number concentration also on meteorological variables, e.g. mixing layer height, wind speed or atmospheric pressure, showing the important influence of regional meteorology on local pollution conditions. Modena can be considered a representative test area of the effect of the meteorological regime for the Central Po Valley on atmospheric particle concentration patterns, characterised by steady high-background concentration. Keywords PM2.5 . Particle number size distribution . Urban background . Po Valley
1 Introduction Atmospheric particles are the object of a rising concern from environmental agencies, public opinion and scientific community. Excessive concentration of airborne particles are known to have detrimental effects on human health (e.g. Oberdörster et al. 2005; Pope and Dockery 2006), on global heat balance (IPCC 2007; Seinfeld 2008) and possibly on precipitation patterns (e.g. Givati and Rosenfeld 2007; Gong et al. 2007). Air-borne particles composition
266
and size distribution are key factors in assessing these effects. Nevertheless, the regulations within the European Union on atmospheric particle concentration involve weight concentration of particles with aerodynamic diameter smaller than 10 μm (PM10, 1999/30/CE) and than 2.5 μm (PM2.5, 2008/50/CE); the prescribed reference methods are based on gravimetric sampler measurement which are quite onerous and associated with a non-negligible uncertainty, whose assessment might be sometimes difficult due to its dependence on operator skill. European directive 96/62/CE requires mitigation actions in case of concentration above limits; Italian local authorities generally adopted limitations to vehicular traffic either on an emission basis or on an area basis. Numerous studies have been presented on highly time-resolved particle number concentration in the Po Valley: Baltensperger et al. (2002) analysed the diurnal pattern of submicron particles at ambient conditions and after conditioning in a urban and rural site near Milan, through the means of a Differential Mobility Particle Sizer (DMPS) and a Hygroscopicity Tandem Differential Mobility Analyser (HTDMA); Van Dingenen et al. (2005) used a HTDMA to have an insight on volatile and hygroscopic properties of submicron particles at the top of Mount Cimone, the highest peak on the Southern side of the Po Valley. Hamed et al. (2007) investigated controls on nucleation events in San Pietro Capofiume, a rural site in the Central Po Valley, from a 3-year time series of DMPS measurements; particle number size distributions in the urban atmosphere of Milan have been measured by Lonati et al. (2006) through an Optical Particle Counter and by Rodriguez et al. (2007) through a DMPS: the former investigated seasonal and diurnal patterns along with traffic patterns for particles within 300 nm–20 μm in diameter, while the latter compared physical–chemical properties of suband super-micron particles among background sites in Barcelona, London and Milan. In this study are presented data collected in an urban background site in Modena, a small-size city in the Central Po River Valley: in 2008, average yearly PM10 concentration has been 39 μg/m3, and EU daily concentration limit has been overpassed in 92 days (ARPA 2008). The first aim of the study is to investigate main removal and accumulation processes in the atmosphere at the sampling site and to assess
Water Air Soil Pollut (2011) 220:265–278
their effect on particle concentration patterns. A further aim is to have a first insight of the PM2.5 profile for the local Po Valley area and to perform a preliminary test to verify if the paradigm proposed by Lenschow et al. (2001) for PM10 regional profile holds for PM2.5, i.e. if rural background PM2.5 is lower than urban background concentration. Within this aim, low particle concentration at the urban background site has been compared with background particle concentration values for the local Po Valley area. The study has involved data in two different season, since it is well established in the literature how seasonality and hourly time resolved data allows to perform a correct exposure assessment (e.g. Bigi and Harrison 2010; Moreno et al. 2009).
2 Sampling Site and Experimental Setup The sampling site is in the grounds of the University campus in Modena, at 34 m a.m.s.l., a city of about 180,000 inhabitants and similar to most of the medium– small cities within the Po Valley (e.g. Bologna, Parma, Reggio-Emilia) which can account for more than 1,200,000 people (ISTAT 2001). The site is not directly affected by the largest metropolitan and industrial area of the valley, which resides in the Milan Province, 180 km North-West; similarly, the direct influence of the Adriatic sea is probably small compared to local sources, being the seaside 120 km East of Modena. The regional meteorological regime is mostly influenced by Alps and Apennines chains bounding the North, West and South sides of the Po Valley. The site is 70 km North of the Mount Cimone Global Atmospheric Watch site (Marinoni et al. 2008); its meteorology is mostly characterised by recurrent wind calms episodes and high-pressure conditions. The university campus is representative of urban background conditions for the city, since no significant source is present nearby, although the site is bounded by three heavily trafficked streets: two are headed towards the city centre and lie 700 m N and 300 m S of the site, respectively; the third bounding street is the city circular road, which lies 250 m SE of the site, dividing the urban area and the university campus from the rural–agricultural area. Meteorological data have been collected at the urban meteorological station of Modena by ARPA Emilia-Romagna
Water Air Soil Pollut (2011) 220:265–278
(Regional Environmental Prevention Agency) and can be considered representative of the local meteorology. Mixing layer depth at the site has been estimated through CALMET simulations by ARPA Emilia-Romagna. The particle data presented in this study have been collected through a Filter Dynamics Measurement System (FDMS, Rupprecht & Patashnick Co.) and Dynamic Mobility Analyser (DMA), at the Department of Mechanical and Civil Engineering of the University of Modena and Reggio-Emilia, inside the university campus. The FDMS device has been inserted in a conventional TEOM previously installed at the measurement site. The system is designed in order to perform a 6-min long conventional TEOM measure and a 6-min long purge of the sampled air on a 47-mm TX40 filter kept at 4°C; the air purged on this latter filter is consequently addressed to the TEOM sensor. The system accounts for possible particulate mass changes occurred during the purge cycle and corrected the previous TEOM measurement (Grover et al. 2005). The sampling inlet, provided of an EPA PM10 impactor and a PM2.5 sharp cut cyclone, is set up on the roof of the department building, 13 m above the ground and 1.5 above the building top. PM2.5 hourly observation records are obtained by this equipment. Sub-micron atmospheric particles have been sampled through the means of a DMA coupled with a Condensation Particle Counter (CPC). The experimental apparatus has been built at the Laboratoire de Météorologie Physique of the University of Clermont-Ferrand, France, and it is composed of a commercial CPC (Model 3010, TSI Inc.) and a DMA column 440 mm long, with an inner radius of 0.196 mm. The instrument employs a recirculating flow system, where excess flow (Qex) is filtered, cooled at atmospheric temperature, dried to 30% relative humidity and used as sheath flow (Qsh). The DMA operates at a Qex =Qsh =5 l/m and at aerosol flow (Qa) = monodisperse sample flow (Qs) = 1 l/m and has been employed in scanning mode as a Scanning Mobility Particle Sizer (SMPS) with a scan time of 120 s for a particle electrical mobility diameter interval of 10–700 nm. Full details of DMA and SMPS can be found in Knutson and Whitby (1975) and Wang and Flagan (1990), respectively. The instrument has been installed at the second floor of the Department of Mechanical and Civil Engineering of Modena. Atmospheric
267
aerosols have been sampled through a stainless steel inlet leaning 2 m outside the window connected to the DMA inlet through a Tygon black tube. Prior the measurements, proper instrument sizing has been calibrated through polystyrene latex spheres. Particle number size distributions (PNSD) and total particle count (N10–700, particles with diameter range 10– 700 nm) are obtained by SMPS. Simultaneous measurements of FDMS and DMA have been analysed in detail in this study, from February 6th to July 31st 2008, consisting in a dataset of about 3500 hourly data for each variable.
3 Data Analysis The monitoring period spans over winter and summer seasons and the study has been focused on hourly, weekly and seasonal variations in particles concentration. The observation period has been therefore split both in a cold period (1600 hourly samples between February 6th–April 30th) and a warm period (1900 hourly samples between May 1st–July 31st) to compare winter and summer trends and in weekdays (Monday through Fridays) and Sundays; Saturdays have not been included in the analysis since their traffic pattern is a mixture of weekdays and Sundays (e.g. Wehner et al. 2002). Data analysis has been performed on observed hourly time series of PM2.5, particle number size distributions and total number concentrations for particles with a diameter range of 10–700 nm. Basic statistics of the particle concentration are presented in Table 1. Sampled aerosol size distributions have been described as the sum of up to three lognormal probability density functions, each one characterised by its integral number concentration (N), geometric mean diameter (Dg) and geometric standard deviation (σg). Size distribution data have been fitted through the software DistFit (version 2009.01, Chimera Technologies Inc.), applying an upper and lower constraint on Dg and σg, i.e. Dg <35 nm for nucleation mode, 35 nm
70 nm for accumulation mode; σg has been constrained between 1.2 and 2.1 for all modes. Generally, winter concentrations are higher both for submicron particle number and PM2.5, as common to other temperate climate sites (e.g. Van Dingenen et al. 2004), because of low vertical mixing, generally
268
Water Air Soil Pollut (2011) 220:265–278
Table 1 Basic statistics of particle concentration measurement Parameter
Winter Min
PM2.5 concentration [μg/m3]
2.9
Summer Median 34.6
Mean 46.5
Max 194
Min 2.1
Median 23.2
Mean 23.3
Max 59.2
Particle count 10–30 nm [1/cm3]
213
8157
10274
56872
108
3929
5155
41773
Particle count 30–50 nm [1/cm3]
394
4825
6578
38367
143
3295
3959
26450
Particle count 50–100 nm [1/cm3]
741
7580
9937
40563
306
3781
4578
25188
3
555
7312
8510
23461
310
3054
3419
13482
Particle count 100–700 nm [1/cm ]
higher emission rates and colder atmospheric temperatures. These circumstances are enhanced by Po Valley morphology, where thermal inversions and low wind speeds are frequent, particularly during winter. 3.1 PM2.5 Diurnal Pattern Boxplots of hourly PM2.5 in Fig. 1, separated between winter and summer, show how generally winter concentrations have higher median and higher variation range than summers. During winter weekdays, median concentration ranges between 25 and 50 μg/m3; this median value increases slightly during morning rush hour and steadily after 1900 Fig. 1 Boxplots for PM2.5 measurements by TEOM-FDMS
hours. Many of the concentration peaks have occurred between February 22nd and March 1st, during conditions of persistent subsidence over the Po Valley. PM2.5 concentration on winter Sundays shows lower variability, besides between 0300 and 0500 hours when median and interquartile range is similar to weekdays: this behaviour is consistent with SMPS measurements and data between 0300 and 0500 hours can be regarded as representative of conditions of atmospheric stability and negligible dust re-suspension. On winter weekdays, daily maximum occurs at 2300 hours, i.e. few hours later than evening rush hour, and it is possibly driven by poor dispersion conditions due to the development of a nocturnal thermal inversion layer. On winter
Water Air Soil Pollut (2011) 220:265–278
Sundays, median concentrations are slightly lower than on weekdays, although with a similar range. Maximum occurs from 0100 to 0300 hours. Similarly to weekdays, mixing layer depth controls daily minimum concentration, which occurs at 1700 hours, i.e. 2 h later than maximum development of mixing layer. During summer, the diurnal cycle for PM2.5 exhibits very narrow range, featured by a minor increase in concentration during morning rush hour. On summer weekdays (when median concentration ranges between 20 and 25 μg/m3), the peak in median concentration occurs at 0900 hours; however, Sundays median concentration is higher than weekdays between 0100 and 1000 hours and is lower than weekdays between 1100 hours and midnight, consistently with local SMPS observations (see Section 3.2). Daily maximum in median PM2.5 on summer Sundays occurs at 0300 hours and is most likely associated to nocturnal stability and night-time traffic, which is probably higher on summer weekends than winter. PM2.5 resulted mostly correlated to 100 nm mode particles as seen in similar sites in the Po Valley, possibly because of condensation processes contributing to an increase in weight concentration (Rodriguez et al. 2007).
Fig. 2 Boxplots for total particle number concentration by SMPS
269
3.2 Particle Number Concentration Diurnal Pattern Total particle count and mean PNSD are remarkably different between winter and summer; in Fig. 2 are shown boxplots of diurnal variation for N10–700 particle concentration and in Fig. 3a–d the average number size distribution along with size distribution modes for winter and summer during weekdays and Sundays. A diurnal pattern is present for total number concentration during winter (Fig. 2), with two peaks on weekdays during rush hours: the former peak occurs at 800 hours, with nucleation mode Dg around 25 nm (Fig. 3a) probably due to vehicular traffic (Harris and Maricq 2001; Morawska et al. 2008), the latter peak starts at 1800 hours, leading to conditions of steady high concentration levels until 2400 hours. On winter Sundays, total number concentration diurnal pattern is characterised by an almost steady median concentration between 0100 and 1000 hours (Fig. 2), whereas concentration peak and highest variability occur during the end of the day, and minimum levels are associated to the full development of the mixing layer. Nucleation mode Dg during night-time on winter weekdays and Sundays (Fig. 3a, b) increases from 10 nm at 1200 hours to
270 Fig. 3 Average number size distribution along with size distribution mode (white dots) by SMPS
Water Air Soil Pollut (2011) 220:265–278
Water Air Soil Pollut (2011) 220:265–278 Fig. 3 (continued)
271
272
20 nm at 2000 hours, probably because of coagulation processes: this process starts at 1500 hours both on winter weekdays and Sundays, although in the latter, the growth rate is smoother because of the lack of evening rush-hour emissions. Both on winter weekdays and Sundays, the drop in nucleation mode geometric mean diameter between 1200 and 1600 hours is probably driven by the decrease in total number concentration leading to a reduction in coagulation; consistently, nucleation mode Dg is smaller on Sundays daytime, since concentrations are lower. During summer, the diurnal pattern of total particle concentration is different than winter. Median concentration is similar between Sundays and weekdays when the mixing layer is highest, whereas on morning and evening rush hours, the median is higher on weekdays and between 0100 and 0500 hours when is higher on Sundays. The concentration range is higher on weekdays throughout the day. The increase in variability on weekdays morning rush hour is due to vehicular traffic, whose signature is visible in Aitken and nucleation particle count peak; notwithstanding these emissions, weekdays median particle concentration is similar to Sundays, probably because of an abrupt change in traffic patterns around June 7th, when the school year is over in Italy, leading to a dilution of vehicular traffic during the latter part of the summer sampling period. After 2000 hours, the drop in mixing layer depth and the development of nocturnal stability or eventually of an inversion layer lead to coagulation of ultrafine particles, occurring both on weekdays and Sundays (Fig. 3c, d). During the end of the day particle count for Aitken mode particles is higher on weekdays than Sundays, probably due to higher emission rates at daytime, then coagulating with the decrease in atmospheric temperature and the drop of the mixing layer. On summer Sundays, total number concentration peaks at early morning (0100–0300 hours), probably due to vehicular traffic, whose signature is suggested by the increase in particle count for nucleation and Aitken modes between 0000 and 0300 hours, associated to an increase in PM2.5 (Fig. 2). The decrease in total particle count starts at 0400 hours, and it might be ascribed to the decrease in night-time traffic rate first and to the rise of the mixing layer (which begins at 0600 hours in summer) later. As commonly observed (e.g. Petäjä et al. 2007; Hussein et al. 2004), the greater the geometric mean
Water Air Soil Pollut (2011) 220:265–278
diameter for the accumulation mode, the lower the number concentration of the accumulation mode particles.
4 Statistical Analysis Statistical methods involved in data analysis were proper of multivariate statistics and have been successfully used in many studies (see Viana et al. 2008 for a review). An analysis of autocorrelation and crosscorrelation of particle number concentration measurements has been performed in order to estimate particles lifetime and possible cyclical behaviour; the analysis involved all data available for weekdays (Mondays through Fridays) during winter and summer. A principal component analysis (PCA) has been used in order to extract the most informative linear combinations of variables, to reduce the dimension of the sample and to discard linear combinations with negligible information content. In this study, the PCA involved all the variables available, besides those with similar information content to avoid multicollinearity (e.g. total number concentration and size segregated number concentration); each variable of the dataset has been rescaled and centred to unitary variance and zero mean, because of scale differences among the variables. The data involved in the statistical analysis are presented in Table 2. All data have been processed
Table 2 Parameters for statistical analysis Parameter
Symbol
Unit
PM2.5 concentration
PM2.5
μg/m3
Particle count 10–30 nm
N10–30
1/cm3
Particle count 30–50 nm
N30–50
1/cm3
Particle count 50–100 nm
N50–100
1/cm3
Particle count 100–700 nm
N100–700
1/cm3
Relative humidity
HR
%
Mixing layer depth
L
m
Atmospheric pressure
p
hPa
Global Radiation
Q
W/m²
Temperature
T
°C
Wind speed
u
m/s
North wind component
uN
–
East wind component
uE
–
Meteorological parameter
Water Air Soil Pollut (2011) 220:265–278
with statistical software R 2.8 (R Development Core Team 2008). 4.1 Autocorrelation Function Analysis The autocorrelation function has been calculated for particle number concentration divided in four diameter ranges (Fig. 4). Autocorrelation function for winter (Fig. 4a) shows a clear semidiurnal pattern for all time series, suggesting that traffic emissions are responsible for particle count concentration peaks, since rush hours are approximately 12 h apart and that concentration levels drop around midday and midnight. Persistence results highest for accumulation mode particles, whose correlation drops less at lag 6: the difference might be ascribed to coagulation processes reducing concentration level for smaller particles and increasing the level for bigger ones. During summer (Fig 4b), all particle ranges show a slight cyclic behaviour with a small increase in the autocorrelation function at 24 h lag. Bigger particles (N100–700) show higher persistence, as proper of the accumulation mode. Also, particles with diameter ranging between Fig. 4 Autocorrelation function for winter (a) and summer (b) particle number concentration
273
10 and 30 nm show a quite high persistence possibly due to their representativeness of the better dispersion conditions in summer and the lower total particle concentration in atmosphere compared to wintertime. No clear photochemically induced nucleation events resulted from cross-correlation function analysis between global radiation and N10–30 during summer (calculated but not here reported) as observed in similar climatic conditions (e.g. Wehner et al. 2003). The autocorrelation function for PM2.5 remains steadily extremely high, showing almost no cyclic behaviour, both in winter and summer. 4.2 Principal Component Analysis PCA allows having an insight on most relevant relations among variables and it is useful to detect main atmospheric processes, excluding all rare or local process (for a discussion on PCA, see Costabile et al. 2009). All the parameters in Table 2 have been included in PCA, wind direction data after split in North and East components (uN and uE, respectively). Since in the literature, there is no standard threshold
274
Water Air Soil Pollut (2011) 220:265–278
for significance of loading, following Wehner and Wiedensohler (2003), |0.25| has been considered as a satisfactory threshold for discriminating variables significantly contributing to the principal component. Results from PCA are presented in Table 3. 4.3 First Principal Component Both during winter and summer, the first principal component splits the dataset in high and low pollution events, which are mostly associated to local meteorological conditions. High positive loadings for particle number concentrations are associated to low loadings for L, Q and T, and a weaker linear correlation results with PM2.5. In summer, PC1 explains 29% of the total variance, and it is characterised by a highest loading for N50– 100, whereas in winter PC1, explaining 39% of total variance, maximum factor loading occurs for N50–100 and N100–700. This difference might be caused by the higher atmospheric temperature and better dispersion in summer, allowing higher concentration of nucleation particles: as shown in Fig. 4b, lifetime of smallest particles is longer in summer. PC1 results influenced mostly by mixing layer depth and by its diurnal cycle, along with the superimposed diurnal cycle of anthropic emissions. Table 3 Loadings and explained variance for the first three principal components from PCA Winter PC1
Summer PC2
PC3
PC1
PC2
PC3
PM2.5
0.20
0.40
0.28
0.21
0.21
−0.53
N10–30
0.30
−0.40
0.06
0.32
0.08
0.50
N30–50
0.35
−0.31
0.16
0.38
0.22
0.35
N50–100
0.39
−0.22
0.13
0.42
0.22
0.11
N100–700
0.39
0.00
0.17
0.38
0.25
−0.21
HR
0.20
0.50
−0.13
0.19
−0.46
−0.24
L
−0.33
−0.18
0.38
−0.35
0.28
0.19
p
0.22
0.03
0.42
0.09
0.33
−0.10
Q
−0.25
−0.13
0.55
−0.30
0.33
0.10
T
−0.33
−0.15
0.04
−0.20
0.47
−0.10
u
−0.16
−0.32
−0.25
−0.19
−0.16
0.32
uN
−0.18
0.18
0.35
−0.19
0.16
−0.26
uE
−0.11
0.28
0.14
−0.16
−0.08
0.01
Variance% 38.8% 16.4% 11.7% 29.3% 21.6%
9.8%
If the original dataset is sorted according to PC1, high values of PC1, corresponding to high concentration of airborne particles, occur preferably during night time in summer and during either morning rush hour or between 2000 and 2400 hours during wintertime; both in summer and winter, night times high weight concentration level occurs later than rush hour maximum and might be a result of poorer dispersion conditions. A total precipitation depth of 40 and 230 mm has been observed during winter and summer monitoring period respectively; however, no significant loadings resulted for the precipitation variable from this and from previous PCA model runs. Therefore, precipitation has been removed from the analysis, leading to an increase in the loadings of the retained variables and in the explained variance by principal components. 4.4 Second Principal Component In winter, PC2 (16% of total variance) shows high loadings for PM2.5, HR and uE and low loadings for u, N10–30 and N30–50. This result indicates that conditions of wind calms are correlated to a rise in relative humidity and PM2.5, concurrently to a decrease in nucleation and Aitken mode particle count; this change in PNSD may be due to coagulation processes, enhanced by high concentration of bigger particles. Differently from PC1, winter PC2 selects pollution events with PNSD with a dominant accumulation mode. During summer, the second principal component explains 22% of total variance and divides in high pollutions events associated to high-pressure conditions and generally fair weather conditions: high positive loadings resulted for N100–700, p, Q and T, while negative correlation resulted for HR. This result indicates the influence of general meteorological conditions on local PNSD, suggesting the dominance of accumulation mode in this PC and a low correlation with PM2.5. 4.5 Third Principal Component Both in winter and summer, PC3 explains 10% of total sample variance, and it has been retained because of its description of different conditions compared to PC1 and PC2. PC3 for the winter dataset
Water Air Soil Pollut (2011) 220:265–278
exhibits high loadings for PM2.5, associated to a high mixing layer depth, pressure and solar radiation and low wind speed. This PC represents conditions of long-lasting regional atmospheric stability which leads to an increase in particle weight concentration, with minor effects on PNSD and total number concentration. The third PC for the summer dataset correlates high concentration levels for N10–30 and N30–50 with low levels of PM2.5 and N100–700. From the dataset, examination results that these latter conditions are not dependant on the hour of the day. PCA allowed identifying possible differences in pollution patterns, with different PM2.5 concentration along with different meteorological conditions and nanoparticle size distribution modes. Although further PCs comprehend up to the 40% of total variance, they have been discarded since the variance explained by each one was almost negligible. The choice of retaining the first two PCs and inspecting the third to investigate particular pollution condition was consistent with the Scree plot and the Kaiser rule (Jackson 1991)
5 Pollution Events Analysis The days with high, mid and low weight concentration (i.e. with hourly PM2.5>80th percentile, 20th < PM2.5 <80th and PM2.5 <20th percentile) have been separately extracted from the dataset and further
275
analysed. These three categories represent pollution events of decreasing intensity. In Figs. 5 and 6 are presented the diurnal pattern for PM2.5 and particle count for the three cases in winter and summer, respectively, together with the corresponding average mixing layer depth. In Figs. 5 and 6, PM2.5 daily pattern is similar in all the three cases, and only the mean concentration is different: from low to high pollution conditions, there is a mean increase in PM of 160% and of 36% in winter and summer, respectively. On the contrary, particle number concentration change depends on the size fraction: N100–700 mean increase is of 68% in winter and of 20% in summer. The increase in the coarser fraction contributes to the increase in PM2.5 and leads to a decrease in N10–30 at midday (5% mean decrease in winter), probably because of enhanced coagulation processes due to the abundance of bigger particles. Regardless of the level of pollution, both in winter and summer, number concentration pattern depends on rush hours, whereas PM2.5 daily pattern results mostly influenced by mixing layer depth instead of intermittent local sources. The main difference between PM2.5 and particle count is that during high pollution events, the former increases both in minimum and maximum concentration, whereas the latter increases mostly in peak concentration. In the morning, peak in PM2.5 is lagged to the N10–30 peak, which can be considered as a marker for vehicular emissions during rush hour: this lag is higher
Fig. 5 Daily pattern during low-, mid- and high-pollution event days, in winter
276
Water Air Soil Pollut (2011) 220:265–278
Fig. 6 Daily pattern during low-, mid- and high-pollution event days, in summer
in winter (up to 3 h) than in summer, probably because of the better dispersion condition. Similarly, also weight concentration minima are lagged to number concentration minima both during early morning and afternoon. Winter morning rush hour occurs during the rising limb of the mixing layer, and the lag of PM2.5 peak is probably due to the concurrent rise of the mixing layer and both the resuspension and the emission of particles by human activities. The daily minimum in PM2.5 occurs when the mixing layer depth is maximum. A drop in PM2.5 concentration occurs also between 0400 and 0700 hours notwithstanding the conditions of atmospheric stability, probably because of the almost negligible human activities and particle remobilisation in those hours. In winter evening, emissions have a longer lifetime compare to mornings, because of the shallower mixing layer height and colder temperature, leading to condensation and coagulation, acting intensively from 2000 to 2200 hours: an increase in PM2.5 might be ascribed also to condensation processes onto existing particles, as observed in a similar site in the Po Valley by Rodriguez et al. (2007), although both the reduced dispersion and the development of a nocturnal inversion layer may lead to the daily PM2.5 peak at 2300 hours. During night time, intense coagulation processes occur, as suggested by the simultaneous decrease in smaller particle concentration and an increase in N50–100 and N100–700.
6 Discussion and Conclusions Particle number size distribution and PM2.5 have been measured at a background site in Modena, a small city in the Po Valley, in 2008; the site can be considered a representative test area of the effect of the meteorological regime for the Central Po Valley on the daily and seasonal patterns of atmospheric particle concentration. Weight concentration levels resulted mostly correlated to local dispersion condition, as described by first principal component in summer and winter (Table 3) and confirmed by PM2.5 trend (Fig. 5 and 6); similarly, weight concentration resulted higher in winter than summer (Fig. 1). Main principal components illustrated how dispersion conditions control particle count and weight concentration, identifying few meteorological conditions leading to different polluting situations: high concentration levels are observed in winter during regional high pressure conditions and fair weather or during periods of wind calms and high relative humidity. In the former case, there is an increase in weight concentration and shift in particle distribution mode to higher diameters; in the latter case, an increase in weight concentration occurs with a moderate effect on distribution mode. Hourly PM2.5 shows a rather narrow daily concentration range, whose pattern seems mostly driven by the diurnal pattern of the mixing layer depth.
Water Air Soil Pollut (2011) 220:265–278
Consistently PM2.5 diurnal pattern is similar for all pollution conditions. The daily patterns for low pollution events (hourly PM2.5 <20th percentile) may be regarded as the urban background concentration daily profiles for summer and winter in the observation site. During low pollution events, minimum hourly concentration results 15.7 and 15.6 μg/m3 in winter and summer, respectively, while mean daily concentration is 26.3 and 19.6 μg/m3 in winter and summer, respectively. Being the minimum hourly concentration strictly similar between seasons, it can be considered as a background concentration for urban aerosols over the sampling period. In order to have a preliminary insight on PM2.5 profile over a wider area, urban low pollution conditions have been compared to daily mean PM2.5 measurements performed at a rural background site 40 km North of Modena, where an atmospheric monitoring station equipped with a gravimetric sampler (SWAM 5A, FAI Instrument, Rome) has been activated in June 20th 2008. Simultaneous weight concentration measurements at the two sites are available only over the period June 20th–July 31st 2008, when mean rural PM2.5 was 16.9 μg/m3 and ranged between 7 and 28 μg/m3. Mean daily concentration at the rural background site can be considered similar to the minimum hourly concentration at the urban site, given the different instrumental apparatus and the short concurrent sampling period at the sites and lower than mean daily concentration at the urban background site. Finally, a PM2.5 concentration of ∼16 μg/m3 can be considered as a possible background concentration representative of the local Po Valley area, over the observation period. On the contrary, particle number concentration resulted correlated both to dispersion conditions and anthropic sources and exhibited a short response time to vehicular emissions. A pattern in particle number distribution mode resulted from the analysis of the most polluted days: during morning rush hour and afternoon in summer and during morning and evening rush hour in winter the mode exhibited a nucleation mode. During rush hour, the nucleation mode can be attributed to the increase in traffic volume occurring along with an increase in total particle count, whereas during summer afternoons the increase in dispersion conditions prevent coagulation removal processes for smaller particles, extending their lifetime. During
277
winter evening coagulation of finer particles lead to higher modal diameter. The results, if viewed altogether, indicate that the vehicular emissions have an immediate effect on sub micron atmospheric particle concentration, while the PM2.5 daily concentration pattern is mainly driven by atmospheric condition that, in the Central Po Valley, determinates accumulation and persistence of the pollutant load with background concentrations that are slightly affected by seasonality. Acknowledgements Alessandro Bigi was supported by the Italian Minister for University and Research under the project PRIN2006. Funds by Emilia-Romagna regional government under the project SIMECH are acknowledged. Authors are thankful to the ARPA-Emilia-Romagna for providing meteorological data and rural PM2.5 data, and to Dr. Carla Barbieri and Dr. Luisa Guerra (ARPA-Modena) for valuable discussions, within the Research Collaboration Agreement “Fine and ultrafine airborne particles monitoring in the Modena District”.
References ARPA. (2008). Air quality network management committee for the Modena province, Air quality in the Modena province. 18th annual report, pp 84 (in Italian). Accessed at: http:// www.arpa.emr.it/cms3/documenti/_cerca_doc/aria/modena/ report_annuali/report%20sintetico%202008/RelazAria2008. pdf; last accessed on Dec 20th 2010 Baltensperger, U., Streit, N., Weingartner, E., Nyeki, S., Prévôt, A. S. H., Van Dingenen, R., et al. (2002). Urban and rural aerosol characterization of summer smog events during the PIPAPO field campaign in Milan, Italy. Journal of Geophysical Research, 107(D22), 1–6. Bigi, A., & Harrison, R. M. (2010). Analysis of the air pollution climate at a central urban background site. Atmospheric Environonment, 16, 2004–2012. Costabile, F., Birmili, W., Klose, S., Tuch, T., Wehner, B., Wiedensohler, A., et al. (2009). Spatio-temporal variability and principal components of the particle number size distribution in an urban atmosphere. Atmospheric Chemistry and Physics, 9, 3163–3195. Givati, A., & Rosenfeld, D. (2007). Possible impacts of anthropogenic aerosols on water resources of the Jordan River and the Sea of Galilee. Water Resources Research, 43, W10419. Gong, D. Y., Ho, C. H., Chen, D., Qian, Y., Choi, Y. S., & Kim, J. (2007). Weekly cycle of aerosol-meteorology interaction over China. Journal of Geophysical Research, D112, 22202. Grover, B. D., Kleinman, M., Eatough, N. L., Eatough, D. J., Hopke, P. K., Long, R. W., et al. (2005). Measurement of total PM2.5 mass (nonvolatile plus semivolatile) with the filter dynamic measurement system tapered element oscillating microbalance monitor. Journal of Geophysical Research, 110(D7), 1–9.
278 Hamed, A., Joutsensaari, J., Mikkonen, S., Sogacheva, L., Dal Maso, M., Kulmala, M., et al. (2007). Nucleation and growth of new particles in Po Valley, Italy. Atmospheric Chemistry and Physics, 7, 355–376. Harris, S. J., & Maricq, M. M. (2001). Signature size distributions for diesel and gasoline engine exhaust particulate matter. Journal of Aerosol Science, 2, 749–764. Hussein, T., Puustinen, A., Aalto, P. P., Mäkelä, J. M., Hämeri, K., & Kulmala, M. (2004). Urban aerosol number size distributions. Atmospheric Chemistry and Physics, 4(2), 391–411. IPCC, Climate Change. (2007). The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press. ISTAT (National Institute of Statistics). (2001). Census of population and housing. Accessed at: http://dawinci.istat. it/MD/index.html, last accessed on June 25th 2010. Jackson, J. E. (1991). A user’s guide to principal components. New York: Wiley. Knutson, E. O., & Whitby, K. T. (1975). Aerosol classification by electric mobility: Apparatus, theory, and applications. Journal of Aerosol Science, 6, 443–451. Lenschow, P., Abraham, H.-J., Kutzner, K., Lutz, M., Preuß, J.-D., & Reichenbächer, W. (2001). Some ideas about the sources of PM10. Atmospheric Environment, 35, 23–33. Lonati, G., & Giugliano, M. (2006). Size distribution of atmospheric particulate matter at traffic exposed sites in the urban area of Milan (Italy). Atmospheric Environment, 40(supp 2), 264–274. Marinoni, A., Cristofanelli, P., Calzolari, F., Roccato, F., Bonafè, U., & Bonasoni, P. (2008). Continuous measurements of aerosol physical parameters at the Mt. Cimone GAW station (2165 m asl, Italy). The Science of the Total Environment, 391(2–3), 241–251. Morawska, L., Ristovski, Z., Jayaratne, E. R., Keogh, D. U., & Ling, X. (2008). Ambient nano and ultrafine particles from motor vehicle emissions: Characteristics, ambient processing and implications on human exposure. Atmospheric Environment, 42, 8113–8138. Moreno, T., Lavín, J., Querol, X., Alastuey, A., Viana, M., & Gibbon, W. (2009). Controls on hourly variations in urban background air pollutant concentrations. Atmospheric Environment, 27, 4178–4186. Oberdörster, G., Oberdörster, E., & Oberdörster, J. (2005). Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles. Environmental Health Perspectives, 113, 823–839.
Water Air Soil Pollut (2011) 220:265–278 Petäjä, T., Kerminen, V. M., Dal Maso, M., Junninen, H., Koponen, I. K., Hussein, T., et al. (2007). Sub-micron atmospheric aerosols in the surroundings of Marseille and Athens: physical characterization and new particle formation. Atmospheric Chemistry and Physics, 7, 2705–2720. Pope, C. A., & Dockery, D. W. (2006). Health effects of fine particulate air pollution: lines that connect. Journal of the Air & Waste Management Association, 56, 709–742. R Development Core Team. (2008). R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. Rodríguez, S., Van Dingenen, R., Putaud, J.-P., Dell"Acqua, A., Pey, J., Querol, X., et al. (2007). A study on the relationship between mass concentrations, chemistry and number size distribution of urban fine aerosols in Milan, Barcelona and London. Atmospheric Chemistry and Physics, 7, 2217–2232. Seinfeld, J. (2008). Black carbon and brown clouds. Nature Geoscience, 1, 15–16. Van Dingenen, R., Raes, F., Putaud, J.-P., Baltensperger, U., Charron, A., Facchini, M. C., et al. (2004). A European aerosol phenomenology-1: Physical characteristics of particulate matter at kerbside, urban, rural and background sites in Europe. Atmospheric Environment, 38, 2561– 2577. Van Dingenen, R., Putaud, J.-P., Martins-Dos Santos, S., & Raes, F. (2005). Physical aerosol properties and their relation to air mass origin at Monte Cimone (Italy) during the first MINATROC campaign. Atmospheric Chemistry and Physics, 5, 2203–2226. Viana, M., Kuhlbusch, T. A. J., Querol, X., Alastuey, A., Harrison, R. M., Hopke, P. K., et al. (2008). Source apportionment of particulate matter in Europe: a review of methods and results. Journal of Aerosol Science, 39, 827– 849. Wang, S. C., & Flagan, R. C. (1990). Scanning electrical mobility spectrometer. Aerosols Science and Technology, 13, 230–240. Wehner, B., & Wiedensohler, A. (2003). Long term measurements of submicrometer urban aerosols: Statistical analysis for correlations with meteorological conditions and trace gases. Atmospheric Chemistry and Physics, 3, 867– 879. Wehner, B., Birmili, W., Gnauk, T., & Wiedensohler, A. (2002). Particle number size distributions in a street canyon and their transformation into the urban-air background: Measurements and a simple model study. Atmospheric Environment, 36, 2215–2223.