Surface Aerosol Optical Properties during High and Low Pollution Periods at an Urban Site in Central China

The aerosol scattering (σsp) and absorption (σap) coefficients and single scattering albedo (SSA) were continuously measured and analyzed at an urban site in Wuhan of central China from November 2014 to July 2017. The average σsp (532 nm), σap (532 nm), and SSA in Wuhan during the study period were 244 ± 212 Mm, 27 ± 17 Mm, and 0.86 ± 0.09, respectively. The aerosol optical properties had pronounced and distinctive diurnal cycles in the Wuhan area. The σsp and σap exhibited the highest values between 06:00 and 08:00 local time (LT) and the lowest values between 14:00 and 16:00 LT, which was mainly due to being coupled with increasing traffic emissions. PM2.5 was a major contributor to the large optical parameters. During pollution periods, the mean σsp (624 Mm) was roughly 3 times that during clean periods (214 Mm), and the σap (58 Mm) was about 2.5 times that during clean periods (24 Mm). The wind speed and direction also strongly affected the aerosol optical properties during the different periods. The high σsp and σap values for the pollution periods were accompanied by calm winds (0–2 m s), from 0° to 45°, whereas the low σsp and σap values of the clean periods were observed with high wind speeds (above 4 m s), from 0° to 90°. Moreover, the σap showed a clear positive correlation (R = 0.412) with the PM2.5 concentrations for the clean periods, whereas the σsp had a good positive correlation (R = 0.406) with the PM2.5 concentrations during the pollution periods.


INTRODUCTION
Aerosols significantly influence global and regional climates by scattering and absorbing solar radiation and by modifying the radiative properties and microphysics of clouds (Yan et al., 2017;Hu et al., 2010).Aerosols can also lead to environmental problems and can adversely affect human health if they enter the respiratory and vascular systems of the human body (Che et al., 2013;Gong et al., 2017).Moreover, the complex behaviors and properties of aerosols and their uneven distributions worldwide create significant challenges and uncertainties for understanding the anthropogenic influences of aerosols on Earth's climate (Singh et al., 2004;Solomon, 2007;Qi et al., 2016).Therefore, it is essential to investigate aerosol optical properties at large scales (van Donkelaar et al., 2010).
Variations in aerosol optical properties have been widely researched at worldwide monitoring stations (Hansen et al., 1984;Haywood et al., 1997;Andreae et al., 2008;He et al., 2009;Jung et al., 2009).Tiwari et al. (2015) studied the interaction between meteorological factors and aerosol optical properties over Delhi during the winter.Lihavainen et al. (2017) observed the aerosol optical properties for a rural background measurement site in western Saudi Arabia.Meanwhile, many observations of aerosol optical characteristics have been reported for polluted cities in China (Chan et al., 1999;Sun et al., 2006;Ma et al., 2011;Fan et al., 2012;Qi et al., 2012;Zhang et al., 2014;Wang et al., 2015a;Zhu et al., 2015;Yan et al., 2016).Xu et al. (2012) studied the aerosol optical properties of Shanghai in the winter and pointed out that the high concentrations of PM 2.5 and black carbon (BC) found would lead to large optical parameters.Yu et al. (2016) investigated the effect of dust on optical properties and the variation of aerosol optical properties over Nanjing.Yan et al. (2017) researched the aerosol optical characteristics at different types of monitoring sites in Shandong and suggested that the aerosol scattering coefficients have a good correlation with the number concentration of accumulation mode particles.However, the number of studies regarding aerosol optical properties over central China is relatively sparse.
Wuhan is one of the largest industrially and agriculturally productive areas in central China and is located on the banks of the Yangtze and Han Rivers.With the continuously increasing numbers of people and vehicles in recent years, this area is challenged by intense air pollution due to industrial emissions and anthropogenic activities (Cheng et al., 2014;Wang et al., 2017).Some research has been conducted concerning Wuhan to study aerosol optical properties over the central region of China.Gong et al. (2015a) investigated the interaction between polluted gas (SO 2 , NO 2 ) and aerosol optical properties in Wuhan.Wang et al. (2015b) analyzed the long-term changes in optical properties in the Wuhan area.Zhang et al. (2017) investigated the aerosol radiative effect in different spectra under haze and high-humidity urban conditions.However, the effect of meteorological factors and the relationship between PM 2.5 concentration and aerosol optical parameters remains uncertain for this region.Therefore, investigating aerosol optical properties under different periods in Wuhan region is also beneficial for better understanding regional pollution and radiative effect in central China.
In this study, a comprehensive study was conducted to analyze the aerosol scattering (σ sp ) and absorption (σ ap ) coefficients and single scattering albedo (SSA) for an urban site over Wuhan in central China from November 2014 to July 2017.To investigate the influence of PM 2.5 concentrations on aerosol optical properties, aerosol optical properties were analyzed statistically during high and low pollution periods.Meteorological parameters, such as the wind and relative humidity (RH), were also investigated during high and low pollution periods.Finally, the interactions between PM 2.5 concentrations and the aerosol optical parameters for the different periods were analyzed via scatterplots.

Study Site
The observation site (114°21ʹE, 30°32ʹN; elevation: 30 m) was located on the roof of the State Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing (LIESMARS) at Wuhan University (Fig. 1) in central China (Liu et al., 2017a).Wuhan is located in the Yangtze River Basin and has a large number of people and vehicles.Moreover, one of China's largest steelmaking plants (the Wuhan Iron and Steel plant) is located in Wuhan (Liu et al., 2017b).The continually increasing human activities and industrial emissions are a huge challenge to the environment over the Wuhan area.Being the largest city in central China, Wuhan is the most representative of urban sites in central China.

Instrumentation and Data
Aerosol total scattering coefficients (σ sp ) were measured by a total-scatter integrating nephelometer at 450, 550, and 700 nm (Model 3563, TSI, USA).This instrument can measure the scattering coefficients simultaneously at three wavelengths and measures the backscattering intensity at the same time.Ambient air was drawn into the sample chamber at a flow rate of 30 L min -1 , and the scattered light at wavelengths of 450, 550, and 700 nm illuminated the sample chamber at an angle from 7° to 170°.Three photomultiplier tubes obtained the total scattering coefficients at 450, 550, and 700 nm at five-minute intervals.The nephelometer was calibrated every three months with filtered air and CO 2 gas.These operations would contribute to approximately 7% error in the data (Gong et al., 2015b).Due to the scattering coefficient and RH have a good correlation under the high RH (> 80%) condition (Xu et al., 2002).The sample chamber of the nephelometer was subjected to constant temperature (20°) and low RH (< 50%).
A seven-wavelength aethalometer (Model AE31, Magee Scientific, USA) was used to observe the mass concentrations of BC and to retrieve the aerosol absorption coefficients (σ ap ).The sampling head continuously drew the air mass through the inlet port, with a flow rate of 5 L min -1 .The air samples were collected on a quartz fiber filter, and the BC was analyzed at the seven wavelengths of 370, 470, 520, 590, 660, 880, and 950 nm. Tiwari et al. (2015) pointed out that BC is regarded as the sole absorbent of sunlight at 880 nm.Therefore, the mass concentrations of BC at 880 nm were exclusively used to determine the aerosol absorption coefficients in this study (Xu et al., 2012).The mass concentrations of BC were provided in units of µg m -3 .The aethalometer was installed in a black room whose RH was maintained between 40 and 50%.
Grimm 180 (Grimm, Germany) is an online environmental particulate monitor that can simultaneously measure the concentrations of PM 10 , PM 2.5 , and PM 1 in the environment.The pump draws ambient air into the sample chamber at a constant flow of 1.20 L min -1.The green laser light generated by the semiconductor laser source illuminates the sample chamber to obtain the concentration of atmospheric particulates.The range of particulate matter monitoring is in the range of 0.1 to 1500 µg m -3 .The Grimm 180 was installed in a black room as well.The Grimm 180 was regularly maintained every three months via cleaning of the sample chamber and the sampling head.
Meteorological parameters, such as wind speed, wind direction, and RH were obtained from an automatic meteorological station (U30-NRC, Onset HOBO, USA) in Wuhan.

Data Processing
According to a previous empirical formula (Xu et al., 2012;Gong et al., 2015b), the σ ap at 532 nm could be calculated by the following: where [BC] represents the mass concentration of BC (µg m -3 ), and the value 8.28 stands for the conversion factor in m 2 g -1 (Yan et al., 2008).Gong et al. (2015b) and Xu et al. (2012) employed this equation to obtain the σ ap for urban sites in Wuhan and Shanghai, respectively.Yan et al. (2008) measured the absorption efficiency factor in Guangzhou city by utilizing linear regression with a correlation coefficient R 2 = 0.92 of σ ap measured by using a photoacoustic spectrometer (PAS) at 532 nm and BC concentrations observed by utilizing the aethalometer at the wavelength of 880 nm.Therefore, this equation was utilized to calculate the absorption coefficient at a wavelength of 532 nm.
The σ sp was obtained from the integrating nephelometer.The three working wavelengths of the integrating nephelometer are 450, 550, and 700 nm.The scattering coefficient, σ sp (550 nm), was employed to calculate the σ ap at 532 nm, according to the following method adopted by Jung et al. (2009).The σ sp at 532 nm is determined by the following: 532 550 (532 nm) (550 nm) where σ sp (550 nm) is the scattering coefficients measured at 550 nm; λ 532nm and λ 550nm represent the wavelengths of 532 nm and 550 nm, respectively; and A is the scattering Ångström exponent, which can be determined via the scattering coefficients at the wavelengths of 450 and 700 nm (Yan et al., 2017): where σ sp (450 nm) and σ sp (700 nm) are the scattering coefficients measured at 450 and 700 nm, respectively; and λ 450nm and λ 700nm represent the wavelengths of 450 nm and 700 nm, respectively.SSA can be obtained using the ratio of the scattering coefficient to the extinction coefficient (Wang et al., 2015).The extinction coefficient is the sum of the scattering and absorption coefficients.The SSA can be expressed as: where σ sp (532 nm) and σ ap (532 nm) are the scattering and absorption coefficients at 532 nm, respectively.

General Features
Box plots of the aerosol optical properties calculated for the entire research period are shown in Fig. 2. The mean σ ap , σ sp , and SSA (standard deviation) from November 1, 2014, to July 31, 2017, were 244 (212) Mm -1 , 27 (17) Mm -1 , and 0.86 (0.09), respectively.As seen in Fig. 2, the σ ap and σ sp have similar daily cycles, whereby the highest values occurred from 06:00 to 08:00 local time (LT) and the lowest values between 14:00 and 16:00 LT.The severe vehicle emissions during the morning rush hour and the low boundary layer height at night increased the concentrations of PM 2.5 (Fig. 2(d)), resulting in the high values of σ sp and σ ap that occurred in the morning and midnight time (Navas-Guzmán et al., 2007).Moreover, the atmospheric boundary layer height increased with increased solar radiation in the afternoon, leading to the decreased concentrations of pollutants and the low values of σ sp and σ ap in the afternoon (Yu et al., 2016).SSA also expressed a single peak day change, with the highest values in the afternoon hours, from 12:00 to 14:00 LT.This daily variation was similar to the results found at other urban and rural stations (Fan et al., 2010;Gong et al., 2015b;Jing et al., 2015;Yu et al., 2016;etc.).
Fig. 3 shows the relative frequency distributions of σ ap , σ sp , and SSA for the entire study period.The relative frequency of σ ap shows a single peak distribution, and most of the σ ap (about 73%) was concentrated in the range from 10 to 40 Mm -1 .A majority of the σ ap was below 500 Mm -1 , with 80% frequency, and the σ sp less than 300 Mm -1 accounted for about 60%.The frequency distributions of SSA were relatively concentrated, and 80% of the SSA ranged from 0.85 to 0.95.The percentages of the SSA more than 0.9 were about 60%, indicating that most of the aerosols in Wuhan were scatter-related particles.
Table 1 shows the statistics of aerosol optical properties observed in other studies.Overall, the averaged σ sp (532 nm) and σ ap (532 nm) in Wuhan were lower than those in the Beijing-Tianjin-Hebei region and the Pearl River Delta region in China, similar with those in other cities.However, it was larger than those in other countries.The mean σ sp of Wuhan (244 Mm -1 ) was less than that of Guangzhou (418 Mm -1 ) and Beijing (360 Mm -1 ) (Andreae et al., 2008;Wu et al., 2014;Jing et al., 2015), but close to the values measured in Xi'an (270 Mm -1 ), Jinan (204 Mm -1 ), and a nonurban Tuoji Island (210 Mm -1 ) site (Zhu et al., 2015;Yan et al., 2017).The σ sp of 107 Mm -1 in Saudi Arabia was only half of that in our experiment (Lihavainen et al., 2017).The averaged σ ap in this study was 27 Mm -1 , which was nearly identical with the 28 Mm -1 obtained in Nanjing (Yu et al., 2016).However, the 91 Mm -1 in Guangzhou and 64 Mm -1 in Beijing were much larger than the averaged σ ap in this study (Andreae et al., 2008;Jing et al., 2015).The σ ap reported for Wuhan was three times that of the mean σ ap observed in coastal sites such as Tuoji Island (8 Mm -1 ) (Yan et al., 2017).The average value of SSA was 0.86 for the research period, close to the 0.88 in Xi'an and 0.89 in Nanjing (Zhu et al., 2015;Yu et al., 2016), but higher than the 0.8 in Jinan from December 2013 to May 2014 (Yan et al., 2017).

Optical Properties during Different Periods
To better understand the aerosol optical properties during different pollution periods, the experimental data were divided into two parts based on the latest air quality standards of China (GB 3095-2012).Due to the complex atmospheric environment of China, a 24-h-averaged PM 2.5 of more than 75.0 µg m -3 indicates the presence of air pollution.The study periods were classified as clean (PM 2.5 < 75 µg m -3 ) and pollution (PM 2.5 > 75 µg m -3 ) periods.
Fig. 4 shows the daily variation in aerosol optical properties during clean and pollution days.The diurnal patterns of the σ ap and σ sp during the pollution periods are different from those during the clean periods.The σ ap and σ sp under pollution conditions increased in the afternoon between 13:00 and 15:00 LT, while the σ ap and σ sp under clean conditions decreased.This is due to the different concentrations of PM 2.5 during the different periods (Fig. 4(d)).During the pollution periods, the concentrations of PM 2.5 continuously increased in the afternoon because of photochemical reactions (Sun et al., 2006).The high concentrations of contaminants led to the high values of σ ap and σ sp .The daily cycle of SSA during pollution periods was similar to that during clean times.The statistics of aerosol optical properties in the different periods are listed in Table 2.During pollution periods, the mean σ sp (624 Mm -1 ) were roughly 3 times that (214 Mm -1 ) of those for clean days, and the 58 Mm -1 of σ ap was about 2.5 times that of the 24 Mm -1 during clean periods.The SSA (0.9) during pollution days was higher than that (0.85) in clean periods.These results indicate that the pollutant concentrations and aerosol optical parameters obviously increased during the pollution periods.
The relative frequency distributions of aerosol optical properties during the different periods are shown in Fig. 5.The distributions of σ ap during the different periods were both single peak distributions.The relative frequency distribution of σ ap was about 93% larger than the 35 Mm -1 during the pollution days, and 92% of the σ ap was less than the 45 Mm -1 during the clean periods.There is a clear threshold value for σ ap of 40 Mm -1 for distinguishing clean and pollution weather.The distributions of σ sp during pollution periods were bimodal, which indicates many types of contaminants were present.Previous studies showed that the contaminants over Wuhan include dust, industrial emissions, and organic aerosols (Wang et al., 2015;Liu et al., 2017).SSA showed single peak distribution for clean days.For SSA, 67% of the values were lower than 0.9, suggesting that absorption-related aerosols were more dominant during clean periods in Wuhan.During pollution   periods, 95% of the SSA values were concentrated between 0.85 and 0.98.In addition, 60% of the SSA values were more than 0.9, indicating that most of the aerosols during the pollution period were scatter-related particles.

Effects of Meteorology on Optical Properties
Because meteorological factors play a crucial role in the diffusion and accumulation of pollutants, a comprehensive study was conducted to analyze the interaction between meteorological parameters and aerosol optical properties.Fig. 6 shows the wind direction frequency and the hourly mean σ ap , σ sp , and SSA for the different periods.The wind direction frequency was expressed as radius, and wind direction as angle degree.The relationships between wind speed and the hourly mean σ ap , σ sp , and SSA are exhibited in As seen in Fig. 6, the most frequent wind directions were from the north and northeast in Wuhan.During the clean days, the σ sp and σ ap values were relatively low, (< 60 Mm -1 and < 600 Mm -1, respectively) and mostly observed under winds from 0° to 90° (Figs. 6(a) and 6(b)).The low σ sp and σ ap values for 0° to 90° were always accompanied by strong winds (> 4 m s -1 ) (Figs. 7(a) and 7(b)).This indicates that the regional transport from 0° to 45° provided a clean air mass that enhanced the diffusion of pollutants (Feng et al., 2007).The SSA values were large for all directions (Fig. 6(c)).Moreover, in the pollution periods in Wuhan, the dominant wind direction was also from 0° to 90°, and the high σ ap (> 60 Mm -1 ) and σ sp (> 600 Mm -1 ) values were mostly observed under winds from 0° to 45° (Figs.6(d) and 6(e)).However, most of the wind from 0° to 45° was close to calm winds (0 to 2 m s -1 ), which indicates that local sources from the region between 0° to 90° largely influenced the σ sp and σ ap values (Cheng et al., 2011;Gao et al., 2011).Noticeably, the σ ap and σ sp values were very high (σ sp > 1000 Mm -1 and σ ap > 70 Mm -1 ) when the winds   were from 25° to 35° at high wind speeds (2-4 m s -1 ) (Figs. 7(d) and 7(e)).This may be due to the regional middle-and long-distance transport from Heinan Province (Tao et al., 2016;Feng et al., 2007).Tao et al. (2013Tao et al. ( , 2014) ) and Liu et al. (2017) pointed out that the local pollution over Wuhan is usually accompanied by large-scale regional pollution.The polluted air mass from the northeast accelerated the accumulation of pollutants, leading to the high σ sp and σ ap values during this period.The relationship between SSA and winds during the pollution days was similar to that in the clean period.The only difference is that the SSA values were larger in the pollution period.The differences in the aerosol optical properties caused by the winds can be explained by the fact that calm winds are not conducive to the spread of pollutants.The accumulation of pollutants resulted in the high σ sp and σ ap values and high PM 2.5 , whereas strong winds enhanced the diffusion of pollutants, leading to the low σ sp and σ ap values.
Fig. 8 shows the relationships between the concentrations of PM 2.5 and BC with changes in the RH during the different periods.Figs.8(a) and 8(b) show the scatter plot for the concentrations of PM 2.5 and BC during the clean and pollution periods, respectively.As seen in Fig. 8, the high concentrations of PM 2.5 also accompanied the high RH.Meanwhile, the concentrations of BC also slightly increased during the pollution days.Fig. 9 shows the relative frequency distributions of RH during the different periods.The relative frequency distribution of RH was about 60% larger than the 75% seen for the pollution period, whereas the 60% of the RH was less than the 75% during the clean periods.This indicates that air pollution days were usually accompanied by high RH.High RH would promote the hygroscopic growth of aerosol particles (Yang et al., 2015), leading to more contaminants.
Overall, the results confirmed that meteorological conditions affected the generation and diffusion of contaminants, which have great relationships with the aerosol optical properties.High wind speed and low RH are conducive to the spread of pollutants (Tao et al., 2014;Yang et al., 2015).The low concentrations of pollutants  have comparatively low hourly mean σ ap , σ sp , and SSA.Moreover, in previous research (Jung et al., 2009;Tao et al., 2014;Yang et al., 2015), it was declared that stagnant weather conditions are beneficial to the formation of haze or fog and result in high concentrations of pollutants because of weak diffusion and mixing.The continuously stagnant weather condition, accompanied by high RH and low wind speed, results in the accumulation of pollutants in the Wuhan area.The high concentrations of contaminants had the high hourly means of the aerosol optical properties.

Relationship between PM 2.5 and Optical Properties
The relationships between the PM 2.5 concentrations and aerosol optical properties during the different periods are displayed in Fig. 10.Figs. 10(a), 10(c), and 10(e) represent the statistical analysis during the clean periods.Meanwhile,Figs. 10(b),10(d),and 10(f) show the statistical analysis during the pollution periods.As shown in Fig. 10(a), the scatter plot between the PM 2.5 concentrations and σ ap exhibits a positive correlation, with a correlation coefficient of R 2 = 0.412 in the clean days, whereas the R 2 coefficient was about 0.0072 during the pollution periods (Fig. 10(b)).This indicates that the concentration of PM 2.5 has a good relationship with absorbent aerosols (such as BC) during the clean periods.Moreover, the PM 2.5 concentration had a good positive correlation with the σ sp during the pollution days, with a correlation coefficient R 2 of about 0.406 (Fig. 10(c)).However, during the clean days, the correlation coefficient R 2 was only 0.24 (Fig. 10(d)).This result shows that most of the contaminants were scattering aerosols during the pollution period, which confirmed the results shown above in Fig. 5. Finally, there was no significant relationship between PM 2.5 concentrations and SSA.
Combined with the above analysis, it can be deduced that the reason for the good correlation between PM 2.5 concentrations and σ sp is the weather conditions, with low RH and high wind speed during the clean periods.The main sources of pollutants over the Wuhan area are industrial emissions and vehicle exhaust (Cheng et al., 2014;Wang et al., 2017).However, low RH could inhibit the watersoluble oxidation of polluted gases such as SO 2 and NO 2 , and high wind speeds are conducive to the spread of air pollutants (Jung et al., 2009;Tao et al., 2014).Therefore, during the clean periods, the main component of the contaminants was absorbent aerosols (such as BC), leading to the high correlation between PM 2.5 concentrations and σ ap during the clean days.Moreover, during the pollution periods, the high RH (Fig. 8) would promote the hygroscopic growth of aerosol particles and lead to the formation of secondary pollutants, increasing the concentrations of scattering aerosols, such as sulfates and nitrates (Chan et al., 1999;Yang et al., 2015).Low surface wind speed promotes the accumulation of pollutants.Therefore, most of the contaminants were scattering aerosols.This is the reason that the PM 2.5 concentration had a good positive correlation with σ sp during the pollution days.

CONCLUSIONS
In this study, we investigated the aerosol optical properties for an urban site in Wuhan from November 2014 to July 2017.The average σ sp (532 nm), σ ap (532 nm), and SSA in Wuhan during the study period were 244 ± 212 Mm -1 , 27 ± 17 Mm -1 , and 0.86 ± 0.09, respectively.The aerosol optical properties had pronounced and distinctive diurnal cycles in the Wuhan area.The σ sp and σ ap exhibited the highest values between 06:00 and 08:00 LT and the lowest values between 14:00 and 16:00 LT.The large vehicle emissions during the morning rush hour and the low boundary layer height at night increased the concentrations of PM 2.5 , resulting in the high values of σ sp and σ ap that occurred in the morning and at midnight.Moreover, the atmospheric boundary layer height increased with the solar radiation in the afternoon, leading to the decreased concentrations of pollutants and the low values of σ sp and σ ap during this time.The SSA also displayed a single peak day change, with the highest values in the afternoon hours, from 12:00 to 14:00 LT.
The concentration of PM 2.5 was a major contributor to the large optical parameters.During the pollution periods, the mean σ sp (624 Mm -1 ) was roughly 3 times that during the clean periods (214 Mm -1 ), and the σ ap (58 Mm -1) was about 2.5 times that during the clean periods (24 Mm -1 ).The high σ sp and σ ap values for the pollution days were accompanied by calm winds (0-2 m s -1 ), from 0° to 45°, whereas the low σ sp and σ ap values during the clean periods were observed with high wind speeds (above 4 m s -1 ), from 0° to 90°.This was due to the continuous stagnant weather state, accompanied by a high RH and low wind speed, which resulted in the accumulation of pollutants in the Wuhan area; these high contaminant concentrations produced high mean values for the aerosol optical properties.By contrast, the high wind speed and low RH were conducive to spreading the pollutants, leading to comparatively low mean values for the σ ap , σ sp , and SSA.Moreover, the σ ap showed a clear positive correlation (R 2 = 0.412) with the PM 2.5 concentrations during the clean days, whereas the σ sp had a good positive correlation (R 2 = 0.406) with the PM 2.5 concentrations during the pollution days.The reason for this is that during the clean periods, the main component of the contaminants was absorbent aerosols (such as BC), but scattering aerosols, such as sulfates and nitrates, generated by the hygroscopic growth of aerosols were more dominant during the pollution days.

Fig. 1 .
Fig. 1.Elevation map of the study area.

Fig. 2 .
Fig. 2. Diurnal variation in hourly averaged aerosol (a) scattering coefficient, (b) absorption coefficient, (c) SSA, and (d) PM 2.5 during the study period.The square symbol inside each box represents the median value; the cross symbol in each box indicate the 1 th and 99 th percentiles of the data; the dash symbols are the maximum and minimum value.The black horizontal line represents the mean of all the data.

Fig. 4 .
Fig. 4. Diurnal patterns in the aerosol (a) scattering coefficient, (b) absorption coefficient, (c) SSA, and (d) PM 2.5 during clean and pollution periods.Black and red lines stand for the hourly means of pollution and clean days, respectively.The error bars represent the standard deviation.

Fig. 7 .
Fig. 7. Figs.6(a)-6(c) and 7(a)-7(c) represent the statistical analyses for the clean periods.Meanwhile, Figs.6(d)-6(f) and 7(d)-7(f) show the statistical analyses for the pollution periods.As seen in Fig.6, the most frequent wind directions were from the north and northeast in Wuhan.During the clean days, the σ sp and σ ap values were relatively low, (< 60 Mm -1 and < 600 Mm -1, respectively) and mostly observed under winds from 0° to90° (Figs.6(a)  and 6(b)).The low σ sp and σ ap values for 0° to 90° were always accompanied by strong winds (> 4 m s -1 ) (Figs. 7(a) and 7(b)).This indicates that the regional transport from 0° to 45° provided a clean air mass that enhanced the diffusion of pollutants(Feng et al., 2007).The SSA values were large for all directions (Fig.6(c)).Moreover, in the pollution periods in Wuhan, the dominant wind direction was also from 0° to 90°, and the high σ ap (> 60 Mm -1 ) and σ sp (> 600 Mm -1 ) values were mostly observed under winds from 0° to 45° (Figs.6(d) and 6(e)).However, most of the wind from 0° to 45° was close to calm winds (0 to 2 m s -1 ), which indicates that local sources from the region between 0° to 90° largely influenced the σ sp and σ ap values(Cheng et al., 2011; Gao  et al., 2011).Noticeably, the σ ap and σ sp values were very high (σ sp > 1000 Mm -1 and σ ap > 70 Mm -1 ) when the winds

Fig. 5 .
Fig. 5. Relative frequency distributions of aerosol optical properties during the different periods.

Fig. 6 .
Fig. 6.Statistical analysis of the absorption (a, d) coefficient, scattering (b, e) coefficient, and SSA (e, f) variations with the wind direction frequency for the clean and pollution periods.(a), (b), and (c) represent the statistical analysis during the clean periods, (d), (e), and (f) show the statistical analysis during the pollution periods.

Fig. 7 .
Fig. 7. Statistical analysis of the absorption (a, d) coefficient, scattering (b, e) coefficient, and SSA (c, f) variations with wind speed for the clean and pollution periods.(a), (b), and (c) represent the statistical analysis during the clean periods, (d), (e), and (f) show the statistical analysis during the pollution periods.The radius represents the wind speed.

Fig. 8 .
Fig. 8. Statistical analysis of the relationships between the concentrations of PM 2.5 and BC with changes in RH during the (a) clean and (b) pollution periods.The color represents the variation in RH.

Fig. 9 .
Fig. 9. Relative frequency distributions of RH during the different periods.

Fig. 10 .
Fig. 10.Statistical analysis of relationships between the PM 2.5 concentration and aerosol optical properties during the different periods.The red line shows the linear regression curve, and the color bar represents the data density.(a), (c), and (e) represent the statistical analysis during the clean periods; (b), (d), and (f) show the statistical analysis during the pollution periods.

Table 1 .
Statistics of aerosol optical properties measured in this study and as reported for other selected monitoring campaigns.

Table 2 .
Statistics of the aerosol optical properties measured in Wuhan during the different periods.