Characteristics of Aerosol Extinction Coefficient in Taipei Metropolitan Atmosphere

The object of this study investigated the correlation of visibility with aerosol physical properties in the urban atmosphere. The field experiments were conducted in four seasons from August 2010 to March 2011 in National Taipei University of Technology (NTUT), located in the downtown of Taipei city. Integrating nephelometer was used to measure the aerosol optical properties and compared with the calculated values based upon the measurements of aerosol concentration and size distribution. The characterization of visibility was influenced by aerosol concentrations and relative humidity. When the relative humidity was above 70%, the particles usually grew into larger size range and changed their optical characteristics. The results showed that the surface area concentration had a great relationship with extinction coefficient, especially for PM2.5. The correlation between particle number or volume concentration and extinction coefficient was complicated. Hence, visibility could be estimated by particle surface area concentration. The conclusions suggested that particle surface area concentration obtained from number concentration combined with size distribution could be used to estimate visibility.


INTRODUCTION
The visibility is a significant index for air quality and weather observation (Waston, 2002;Chang et al., 2009;Che et al., 2009;Carbone et al., 2013).Previous studies showed that the urbanization and industrialization conduced the air pollution emission increased and the visibility decreased (Qiu and Yang, 2000;Fan et al., 2005).Tsai (2005) found that the average length of visibility was 12.4 ± 4.2 km from 1961 to 2003 in Tainan, Taiwan.However, the average length of visibility from 1997 to 2003 was 8.1 ± 1.4 km.The visibility loss was influenced by air pollutions which were from vehicular emission and fossil-fuel burning (Goyal and Sidhartha, 2003;Tsai, 2005).In China, it was over 200 days a year that the visibility was less than 10 km.Besides, the visibility decreased about 2.1 km every ten years from 1990 to 2005 which was also affected by the urbanization and industrialization (Zhang et al., 2004;Che et al., 2007;Chen et al., 2012).Some researches reported that the atmospheric visibility was associated with PM 2.5 and PM 10 (Pillai et al., 2002;Fajardo et al., 2013;Pui et al., 2014) and also found that the visibility was decreased with increasing aerosol mass concentration (De et al., 2005;Wild et al., 2005;Tiwari et al., 2011;Chen et al., 2012;Liu et al., 2013).Kim et al. (2006) had pointed out that the visibility was affected by fine particle (PM 2.5 ) emission directly or indirectly.The visibility declined from 61.7 to 1.9 km, when the PM 10 concentration increased from 32.9 to 601.8 μg cm -3 during the spring of 2000 in Korea (Kim et al., 2001).Therefore, the atmospheric particles affect not only human health, but also visibility.
Visibility is a quantity that refers to the clarity with which distant targets can see via the visible light.The visible light is a visually observable electromagnetic radiation, and the wavelength range is from 380 nm to 770 nm.In addition, the motion of light is along a straight line.According to the Beer-Lambert-Bouguer law, the visibility was related to the visible light intensity (flux).However, the light absorption and scattering of aerosol would reduce the intensity of the incident light.Therefore, the light absorption and scattering of aerosol was an important parameter in visibility estimation.In particular, the aerosol scattering is associated with the atmosphere visibility significantly (Hinds, 1999;Lee et al., 2005;Liu et al., 2014).Dzubay et al. (1981) compared the measurement aerosol light scattering with the calculation extinction coefficients.The result showed that the scattering effect of particles had the significantly related to the visibility.During the past twenty years, several studies investigated the relationship between PM 2.5 concentration and particle light scattering.The results estimated the light scattering by using fine and coarse particle mass concentration.(Chow and Watson, 2002;Liang et al., 2013).
Furthermore, the visibility is not only influenced by particle properties, but also affected by relative humidity, temperature and other gaseous pollutants, etc. (Malm et al., 2000;Nessler et al., 2005;Yang et al., 2007;Cheng et al., 2008b;Liu et al., 2008;Pan et al., 2009).Especially, high relative humidity could significantly increase the effect of particles on visibility (Malm et al., 2000;Malm and Day, 2001;Liu et al., 2012;Chen et al., 2014;Qu et al., 2015).Some studies showed that the hygroscopic growth of particles affected the optical properties and size distribution, which might have an important impacts on visibility (Wang et al., 2008;Zieger et al., 2013).The water content of particle affected the aerosol properties, like extinction coefficient, cross-section, components and mass concentration etc. (Jung et al., 2009a;Cheng et al., 2011;Sun et al., 2013).The extinction coefficient of the particle is the sum of its scattering and absorption coefficient (Hinds, 1999).The particle extinction coefficient increased with increasing sulfate and nitrate concentrations when the relative humidity was high (Zhou et al., 2014;Qu et al., 2015).
There are three methods for measuring visibility, according to the USEPA Visibility Monitoring Guidance: (1) to collect and analyze the physical properties of atmospheric particles.
(2) to measure light extinction using transmissometer and nephelometer and, (3) to understand the relationship between air pollutants and vision by image analysis.
In the previous study, visibility was mainly obtained by measuring the ratio of light intensity which was traversing the atmospheric aerosol directly, and compared with the particle mass concentration via monitoring.However, there were few information regarding the particle number concentration.The object of this paper focused on investigating the correlation of visibility with aerosol physical properties in Taipei city.

Experimental Sites
The field experiments were conducted in four seasons from August 2010 to March 2011 in National Taipei University of Technology (NTUT), located in the downtown of Taipei city, as shown in Fig. 1.The sampling site was located on the roof of a four-stories building.It is also near the heavy traffic location and important business zone.Traffic emission is a notable source of pollutants.
Taipei is located in the north of Taiwan.The weather in four seasons is significantly different.Summers are hot and humid, accompanied by sporadically heavy rainstorms and typhoons, while winters are generally warm and humid

Instrumentation and Measurements
In the experiment, the optical and physical properties of atmospheric particles were measured continuously for 24 hours each season, included Integrating Nephlometer and outdoor air Electrical Low Pressure Impactor (outdoor air ELPI).The Integrating Nephlometer (TSI, Model 3563) was used to measure scattering coefficient of the ambient atmosphere.The detection limit of light scattering coefficient was about 1.0 × 10 -7 m -1 in blue and green wavelengths.Besides, the scattering angle range of measurement was from 7° to 170°.The total (7° to 170°) and backscatter (90° to 170° signals use a rotating backscatter shutter to block the illuminated sample volume from 7° to 90°.The outdoor air ELPI (Dekati Ltd.) was used to measure the airborne particle size distribution and concentration in real-time.The size range of particle determine was from 7 nm to 10 µm.Continually signals of measured current were converted to the aerodynamic size distribution using particle size dependent relations describing the properties of the charger and the impactor stages (Lin et al., 2010;Syu et al., 2014).Because relative humidity was an important properties in this study.Therefore, the temperature and relative humidity condition were not controlled as the flow into ELPI.In order to analyze qualitative and quantitative atmospheric particles, particle density plays an important role in the parameter that the Stokes diameter of the particle converts to an aerodynamic diameter.Particle density can be obtained by particle geometric diameter and particle mass.In addition, McMurry et al. (2002) found that the average effective density of particles is 1.61 g cm -3 in the urban atmosphere.Due to the reference, the effective density of atmospheric particle was 1.61 g cm -3 in this study.
The numerical analysis was introduced in the following section which included the extinction coefficient, surface area and volume concentrations.The extinction coefficient (σ ext ) was the sum of scattering coefficient (σ sg and σ sp ) and absorption coefficient (σ ag and σ ap ) of gas and particle (Eq.( 1)) and the visual range (L v , visibility) was calculated by Eq. ( 2) (Middleton, 1958): (1) 3.912 The visual range or visibility is inversely proportional to the extinction coefficient.In addition, Hinds (1999) noted that the extinction coefficient is affected by the particle number concentration, cross-section area and extinction efficiency, as shown in Eq. (3).
where N i is particle number concentration; A pi is the crosssection area; Q ei is particle extinction efficiency; and d i is particle diameter.
Besides, the particle volume concentration was calculated by Eq. ( 4): where V is the total volume concentration.
And the particle surface area concentration was calculated by Eq. ( 5): where S is the total surface area concentration.

RESULTS AND DISCUSSION
In this section, all particle properties were compared with the calculated extinction coefficient.Therefore, the relation between calculated extinction coefficient and measured light scattering coefficient were contrasted first before the official discussion.As shown in Fig. 2, the results indicated that the light scattering coefficient had a great relationship with the extinction coefficient.

Aerosol Properties and Relative Humidity
As demonstrated in Fig. 3, time series analysis was used to investigate the aerosol, weather and temporal data properties, which included the number concentration, volume concentration, surface area concentration and extinction coefficient of aerosol, and relative humidity from the ambient atmosphere of four seasons (August as summer, November as autumn, February as winter, and March as early spring) in NTUT sampling site in Taipei.The particle number concentration was decreased as the relative humidity   increased.However, it was quite obvious that the particle surface area concentration increased as the relative humidity was above 70%, compared with the drier condition.It specified that the particles grew when the relative humidity was over 70%.Similar phenomena were found in different seasons.This trend was also reported by Chen et al. (2014).According to the research by Martin and Finlay (2005), condensation process was the change of the physical state of water from gas phase into liquid phase.And the particle played the role of the condensation nuclei.Therefore, the particle grew up in this procedure.Previous studies also indicated that the aerosol grew up as the relative humidity was above 80% (Tsai, 2005;Zieger et al., 2013).
It was also found that the extinction coefficient increased with increasing particle surface area concentration.According to the Eq. ( 3), the extinction coefficient is proportional to the particle cross-section area.In addition, the surface area was four times than the cross-section area of sphere particle.Therefore, this equation could explain the experimental result.
The Fig. 4 indicated the extinction coefficient of particle versus relative humidity for different seasons.Actually, most conditions of particles were nonspherical, and the elements of particle were complicated.Therefore, the moisture content efficiency was unpredictable.However, the results reported that the extinction coefficient increased as the relative humidity increased although it's rough.These results were similar to that results obtained by Liu et al. (2012) and Sun et al. (2013).Consequently, it was interesting to note that the atmospheric visibility converted hazy condition with the particle cross section was increased as in the high relative humidity of ambient atmosphere (Day and Maln, 2001;Stock et al., 2011).

The Correlation between the Extinction Coefficients and the Particle Concentrations (Number, Volume and Surface Area)
Fig. 5 to Fig. 7 proved the correlation between the extinction coefficients and particle concentrations (number, volume and surface area) in different size ranges, Basing on Eq. ( 3), the extinction coefficients could be translated by number concentration, cross-section area and extinction efficiency of aerosol.Hence, the following paragraphs investigated the correlations of extinction coefficients and particle number, surface area and volume concentrations, respectively.The Fig. 5 revealed the correlations of extinction coefficient and the number concentration of PM 2.5 and PM 10 particle.As shown in figure, the number concentration just had a good agreement with the extinction coefficient in summer, compared with the other three seasons.Moreover, the average extinction coefficient in summer was significantly lower than that in other seasons due to the lower particle number concentration.The particle size distribution and number concentration changed owing to the condensation of particle in higher average relative humidity condition.Furthermore, this result was suggested by Shimizu et al. (2015) that the extinction and the number concentration of aerosol had good agreement during the dry season.Yan et al. (2004) also had reported that the extinction of single and larger aerosol is greater than small aerosol.However, it couldn't exclude the other complex physical properties of aerosol that might change the size distribution and number concentration of particle too, like coagulation and deposition.Some references claimed that the particle number concentration declined with increasing aerosol particle size, due to coagulation and condensation (Hinds, 1999;Jung et al., 2006).But, it still got a positive relation with extinction coefficient.
As demonstrated in Fig. 6, the correlations between extinction coefficients and volume concentrations of PM 2.5 was better than PM 10 .According to Eq. ( 4), the number concentration of fine particle was extremely more than coarse particle at the same volume concentration.Besides, basing on the Eq. ( 3), the cross-section area and number     concentration were the important factors of extinction coefficient.However, the volume is the quantity of threedimensional space.It multiplied one more length unit than the cross-section.Therefore, the correlations between extinction coefficient and volume concentration of coarse particle was underestimated then fine particle.In addition, the correlation trend of PM 2.5 and PM 10 for volume concentration was different from the trend for number concentration.This result suggested that the determination of visibility might be confused by using the volume concentration of particles.
As shown in Fig. 7, it was exhibited that the correlations were good between extinction coefficient and surface area concentration of PM 2.5 and PM 10 .In addition, the surface area and cross-section area of sphere particle were also the quantity that expresses the extent of a two-dimension unit.Therefore, according to this contention and Eq. ( 3), the extinction coefficient was affected by particle surface area concentrations obviously.As said by previous studies, the light intensity would be reduced by light scattering and absorption characteristic of particle.In atmospheric condition, the primary ingredient of the extinction coefficient was the light scattering of particles, particularly (Hinds, 1999;Lee et al., 2005).Consequently, the particle surface area concentrations which obtained from number concentration combined with size distribution c could be used to determine the visibility.

Cumulative Percentage of Particle Properties
The cumulative percentages of particle number, surface area, volume concentration and extinction coefficient were demonstrated in Fig. 8.The cumulative percentage of PM 2.5 number concentration and volume concentration was more than 99% and less than 80%, respectively.However, the cumulative percentage of extinction coefficient was from about 80% to 97%, which was partially unconnected with particle geometric properties except the surface area.Besides, the cumulative percentage of PM 2.5 extinction coefficient in summer was lower than other seasons.It suggested that less fine particles suspended in the atmosphere.
Furthermore, the correlation between particle volume concentrations and extinction coefficients was good as the particle size was less than 0.34 μm.This might be due to Rayleigh scattering of smaller particles and the extinction efficiency was not as that of micrometer-sized particles valued in certain range (Hinds, 1999;Friedlander, 2000).However, the surface area concentration was highly correlated to the extinction coefficient as the particle size was greater than 0.8 µm.

CONCLUSIONS
This research presented the aerosol measurements and optical property analysis sampled in Taipei city from August 2010 to March 2011.The results indicated that the particles might grow into large size range when the relative humidity was greater than 70%.The volume concentration increased with increasing the relative humidity.However, the particle number concentration did not significantly change since the particle growth was mainly formed by humidity condensation.Particle surface area concentration was found the main factor to affect the extinction coefficient in this study.The surface area concentration was highly correlated to the extinction coefficient as the particle size was greater than 0.8 µm.However, the particle volume concentration was related to the extinction coefficient as the particle size was less than 0.34 µm.The results showed that the surface area concentration had a great relationship with extinction coefficient, especially for PM 2.5 .Therefore, the particle surface area concentration could be used to estimate the atmospheric visibility.

Fig. 1 .
Fig. 1.Location of the sampling site at NTUT in Taipei City, Taiwan.

Fig. 2 .
Fig. 2. The correlation between the extinction coefficients and the scattering coefficients.

Fig. 3 .
Fig. 3. Time series of the aerosol concentration, extinction coefficient and relative humidity in Taipei City for different seasons: (a) summer, (b) autumn, (c) winter and (d) spring.

Fig. 5 .
Fig. 5.The correlation between the extinction coefficients and the particle number concentrations in different size ranges.

Fig. 6 .
Fig. 6.The correlation between the extinction coefficients and the particle volume concentrations in different size ranges.

Fig. 7 .
Fig. 7.The correlation between the extinction coefficients and the particle surface area concentrations in different size ranges.

Fig. 8 .
Fig. 8.Comparison of cumulative particle percentage and cumulative percentage of extinction coefficient was between the seasons.