P ROCESOS DE ELABORACIÓN

The daily variation of the total particle number concentration (D p_14nm−660nm) and the total particle number concentration (3 nm to 2.5µm (CN3)) over the period of the campaign are shown in Fig. 3.3 (a)). There was evidence of morning, afternoon and night peaks in both PNC_14nm−660nmand CN₃. The morning peak was around 6:00 to 7:00 AEST, while the afternoon and night peaks were observed from around 12 noon to 14:00 AEST and 22:00 to 23:00 AEST, respectively. The greatest contributor to particle number size distribution ranged from 14 nm to 660 nm over the measurement period was ultrafine particle (< 100 nm) (Fig. 3.3 (b)). Bivariate polar plot shows the relationship between wind direction and wind speed (Fig. 3.3 (c)). During the campaign, high particle number concentrations were observed in the south and northeast sectors at high wind speed. Under low wind speed conditions, high particle number concentrations were observed in the south west and southeast. Factors that contributed to this observation are discussed in Section 3.5.

Fig. 3.3: Total particle number concentration (measured from 16^th January 2013 to 15^thFebruary 2013; (a) average diurnal variation of total particle number concentration (PNC_14nm−660nm) and total particle number concentration range from 3 nm to 2.5 µm (CN3) with the 95 % confidence level shaded, (b) average particle number size distributions from 14 nm to 660 nm and (c) dependence of total particle number concentration (PNC_14nm−660nm) on wind speed and wind direction. Wind speed and wind direction are represented by the concentric circles and coloured shape, respectively. Average particles counts are illustrated by the colour bar. Note that 10-min averaged data were used.

Distribution

3.3.1 PM

The daily mean mass concentration of PM_2.5from 24^thJanuary 2013 to 15^thFebruary 2013 during the aerosol phase of the MUMBA campaign was 6.1 µg m⁻³. The maximum daily average was 22.1µg m⁻³(Fig. 3.4). The daily mean PM_2.5 measured at the nearby Office of Environment and Heritage (OEH) for the same time period was 6.2µg m⁻³. The OEH station is located approximately 3 km from the MUMBA campaign site (Fig. 2.1(b)). This suggests that the two sites were exposed to similar sources. The daily mean of PM2.5 obtained in this study was always below the maximum allowable daily mean as specified by the Australian National Environment Protection (Ambient Air Quality) Measure (Air NEPM) of 25 µg m⁻³ (not to be exceeded) and by the United States Environmental Protection Agency, US EPA (35 µg m⁻³, 98^th percentile, averaged over 3 years).

Fig. 3.4: Time series of daily average mass concentration of PM_2.5measured from 24^thJanuary 2013 to 15^thFebruary 2013. The blue dashed line is daily mean mass concentration of PM_2.5. The black dashed line is the maximum allowable daily mean as specified in the Australian National Environment Protection (Ambient Air Quality) Measure (Air NEPM) which is (25µg m⁻³).

3.3.2 Characteristics of Particle Number Concentration

Particle size distribution is an important way of understanding aerosol properties and their sources [Charron et al., 2008; Harrison et al., 2011]. While fine particle (particle with diameter < 2.5 µm) are considered to be of a significant concern for health, there is also an emerging focus on ultrafine particles (particle with a diameter <100 nm) [Harrison and Yin, 2000; WHO et al., 2013]. The focus of this study is on PNC_14nm−660nm, CN₃ and particle number ranged from 3 nm to 100 nm (PNC_3nm−100nm). Figure 3.5 (a) illustrates the time series of the 10-minute averaged particle number concentration of PNC_14nm−660nm and CN₃. Number concentration of PNC_14nm−660nm ranged between 3.6 x 10² cm⁻³ and 1.8 x 10⁵ cm⁻³, while CN₃

ranged from 4.6 x 10² cm⁻³ to 6.5 x 10⁴ cm⁻³.

An extreme event was observed on the 16^th of January 2013 (Fig. 3.5 (a)) at approximately 22:00 (Fig. 3.5 (b)). The 10-minute averaged particle number concentration of PNC_14nm−660nm at 22:00 was 1.8 x 10⁵ cm⁻³. The monitoring site experienced increase in wind speed between 18:30 and 21:00 (Fig. 3.5 (d)) and wind blowing from the southwest sector (Fig. 3.5 (c)). After 21:00, wind speed started to ease and this could have reduced the dilution of particle number number concentration which then resulted in the extreme event.

Fig. 3.5: (a) Time series of PNC14nm−660nm and total particle number concen-tration range from 3 nm to 2.5µm (CN3) from 16^thJanuary 2013 to 15^thFebruary 2013 and (b) time series of PNC_14nm−660nmon 16^th January 2013. Figure (c) and (d) are wind direction and wind speed on 16^th January 2013, respectively. The rectangular orange box delineates the extreme event. Note that 10-min averaged data was used.

The mean of CN₃ over the period of 16^th January 2013 to 15^th February 2013 was 7.4 x 10³ cm⁻³ with a median value of 5.5 x 10³ cm⁻³. During the aerosol measurement period, aerosol populations were dominated by the ultrafine particles (particle with a diameter <100 nm) as shown in Fig. 3.3(b). On that account, there are a few assumptions made in this work including particle numbers with size diameter ranging from 660 nm to 2.5 µm nm are negligible. Another assumption was CN₃ provides information on ultrafine particles (PNC_3nm−100nm). Therefore, the ultrafine particles concentrations in this study were calculated using Eq. 3.1 to 3.3. Particles from 3 nm to 14 nm (abbreviated as “a_3nm−14nm”) were obtained by subtracting PNC_14nm−660nm from total CN₃. Then “a_3nm−14nm” was added to the

Distribution

sum of particle number from 14 nm to 100 nm (abbreviated as“b_14nm−100nm”) to obtain PNC between 3 nm and 100 nm (abbreviated as “c3nm−100nm”).

a_3nm−14nm= Total CN₃− PNC_14nm−660nm (3.1)

b14nm−100nm= Sum of PNC from PNC14nm to PNC100nm (3.2)

c_3nm−100nm = a_3nm−14nm+ b_14nm−100nm (3.3)

The observation of particle concentration over the size range of 14 nm to 660 nm measured by the SMPS in this study was compared with other studies at coastal environments (Table 3.1). The mean and median of the 10-minute averaged particle number concentration between 14 nm and 660 nm measured during the aerosol measurements period were 5.2 x 10³ cm⁻³ and 3.1 x 10³ cm⁻³, respectively. The mean value reported in this study was comparable to studies by Peng et al. [2014]

on a particle number ranging from 15 nm to 660 nm at a sub-urban coastal site in China and by Sorribas et al. [2011] on particle number between 14 nm and 673 nm at a coastal-rural environment in Spain. However, the mean value of particle number ranging from 14 nm to 660 nm in this study was lower than the study by Pey et al. [2008] on particle number ranging from 13 nm to 800 nm at a coastal-urban environment in Barcelona, with a mean value of 16.9 x 10³ cm⁻³. One can conclude that the particle number concentrations observed during this campaign fell within range of the particle number concentration observed elsewhere in marine and urban environments.

Table 3.1: Comparison between this study and other studies. PNC is particle number concentration. NA is not available.

PNC x 10³(cm⁻³)

Study Site Size (nm) Mean Median Quartile, Q1

(25% percentile)

Quartile, Q3 (75% percentile)

This study Urban-marine 3-100 7.0 5.2 2.7 9.5

This study Urban-marine 14-660 5.2 3.1 1.6 6.2

This study Urban-marine 3-2500 7.4 5.5 3.0 9.9

Cheung et al.(2011) Urban 4-110 9.3 NA NA NA

Peng et al. (2014) Coastal 15-660 5.7 3.9 NA NA

Sorribas et al. (2011) Coastal-rural 14-673 8.7 7.1 NA NA

The mean and median of ultrafine particles (PNC_3nm−100nm) over the period of 16^th January 2013 to 15^th February 2013 were 7.0 x 10³ cm⁻³ and 5.2 x 10³ cm⁻³, respectively which is comparable to the value reported by Cheung et al. [2011] for a sub-tropical urban environment in Australia (Table 3.1). The calculated mean total CN₃ and PNC_3nm−100nm shows that the particle population observed in this study was dominated by ultrafine particles.

3.3.3 Classification of Particle Number Concentration

Principal component analysis (PCA) (details in Section 2.2.3) was performed on the total PNC_14nm−660nm and the distribution of component loadings is illustrated in Fig. 3.6. Component loadings (factor loadings), reflect the correlation between the factors and variables (particle population in the individual size bins). In this study, strong component loadings (value ≥ 0.75, [Liu et al., 2003] as dashed line in Figure 3.6) were chosen for result interpretation. That is, particle numbers were summed across the size ranges “above the dashed line” and identified as Factor 1, Factor 2 and Factor 3 (Fig. 3.6).

Fig. 3.6: Distribution of component loadings on the entire 10-minute averaged particle number concentration. Only component loadings that were equal to and greater than 0.75 were used to interpret the results. Note: x-axis is in log scale.

Given their distribution, Factor 1 is labelled as the Small Factor (N_S) (14 nm

< Dp < 50 nm), Factor 2 is labelled as the Medium Factor (N_M) (60 nm < Dp <

150 nm) and Factor 3 is labelled as the Large Factor (N_L) (210 nm < Dp < 450 nm). Factor 1 and Factor 2 described 31% of the variability, respectively meanwhile, Factor 3 described 27% and all three size fractions described 89% of the cumulative variance. The Small, Medium and Large factors used in this study are approximately equivalent to Nucleation mode, Aitken mode and Accumulation mode, respectively.

It is worth mentioning that Nucleation mode sizes (D p < 10 nm) were actually not within the dataset used in this study (PNC_14nm−660nm). The temporal variations in these factors are discussed in Section 3.5.