Size Distribution and Optical Properties of Particulate Matter (PM10) and Black Carbon (BC) during Dust Storms and Local Air Pollution Events across a Loess Plateau Site

We analyzed the suspended particle size distribution in the range of 0.5 to 10 μm and the optical properties of the particles from March 2007 to December 2010 at a site on the Loess Plateau (SACOL; 35.57°N, 104.08°E; 1965.8 m a.s.l.) about 48 km southeast of the center of Lanzhou. The results indicated that the variation in PM10 was much larger in spring than in winter because of frequent dust events or local blowing soil dust during spring. The highest number concentrations of coarse-mode particles were likely attributable to dust events that transported mineral dust or soil dust in the spring season, caused by cold fronts or strong local winds. In contrast, the fine-mode particles that dominated in the cold season at SACOL were probably indicative of anthropogenic sources related to fossil-fuel combustion and biomass burning. The comparison of dust events and anthropogenic air pollution shows a clear distinction of lower PM10 with higher Bap for pollution episodes and higher PM10 with lower Bap for dust events. These findings suggest that the results in the cold season were likely attributable to light absorption of black carbon, and the coarse mode particles were dominant during dust events in spring.


INTRODUCTION
Tropospheric particulate matter (PM 10 ) and black carbon (BC) play key roles in regional and global climate depending strongly on their physical and optical properties (Ramanathan et al., 2001b;Arimoto et al., 2006;Li et al., 2011;IPCC, 2013;Bi et al., 2014).They are very effective at modifying the radiation field by absorbing and scattering solar and thermal radiation (Anderson et al., 2003;Huang et al., 2006a, b;Li et al., 2007;Ramanathan and Carmichael, 2008;Bi et al., 2010;Wang et al., 2010) and altering the precipitation rate and hydrological cycle (Rosenfeld et al., 2001;Ramanathan et al., 2005;Su et al., 2008).The Loess Plateau is located in northwestern China, near the largest and most persistent sources of airborne dust in the world, including the Taklimakan, Gobi, Badain Jaran, and Tengger deserts (Zhang et al., 2003a, b;Xu et al., 2004;Sun et al., 2005;Wang et al., 2008;Igarashi et al., 2011;Che et al., 2013).Dust aerosol, one of the major aerosol types of PM 10 , is an active component in the climate system (Prospero, 1999;Zhao et al., 2008).Driven by midlatitude prevailing westerlies, large quantities of mineral particles can be lifted to high altitudes during dust events and thus transported to the eastern Pacific and North America (Jaffe et al., 1999;Prospero, 1999;VanCuren and Cahill, 2002;Huang et al., 2008a;Prospero and Mayol-Bracero, 2013).Additionally, the region also has major sources of anthropogenic aerosols produced by human activities (Chen et al., 2014;Huang et al., 2014).For instance, BC, which originates from incomplete combustion, is an anthropogenic substance capable of damaging human health and leading to serious climatic effects at scales from regional to global (Ramanathan and Carmichael, 2008;Huang et al., 2011;Wang et al., 2013;Zhang et al., 2013).It is also the strongest absorber of solar radiation, thus contributing to warming of the atmosphere.Additionally, BC affects climate by indirect effects due to its action on cloud droplets and cloud microphysical properties.When BC mixes with dust aerosols during dust events, it significantly changes the aerosol physical and optical properties (Ramanathan et al., 2001a;McKendry et al., 2008;Fu et al., 2009;Zhou et al., 2013;Wang et al., 2015).Therefore, understanding the physical processes and optical properties of aerosol particulates across the Loess Plateau is important (Tegen et al., 1996;Liu et al., 2011).

Site Description
The remote observation site SACOL (35.57°N, 104.08°E, 1965.8 m a.s.l.) has been in operation since 2005 (Huang et al., 2008b;Wang et al., 2015).SACOL is located on the Loess Plateau at the western edge of the Hexi Corridor in Gansu Province (south of the Yellow River), about 48 km southeast of central Lanzhou.The motivation for establishing this site was to develop a deeper understanding of the physical and optical properties of PM 10 and BC in the semi-arid and arid regions near the border of the Loess Plateau and to investigate the impacts of natural and anthropogenic pollution on the regional climate (An et al., 2007;Costabile et al., 2010;Zhang and Li, 2011;Zheng et al., 2013).The characteristic ecosystems at SACOL include various types of vegetation, such as needle-leaf tree, short grass, meadow, marsh/desert, marsh/semi-desert, irrigated crop, and glacier/snow.

Instrumentation
Since March 2007, the mass concentration of PM 10 has been measured continuously by an R&P1400a analyzer based on the principle of the tapered element oscillating microbalance (TEOM) with the appropriate sample inlet.The instrument works at 50°C in order to dry the aerosols, with a flow rate of 16.7 L min -1 .Aerosol scattering coefficient measurement was performed using single-wavelength ECOTECH integrating nephelometers (M9003).The nephelometer was available at the wavelengths of 450, 520, and 700 nm and was designed specifically for studies of the direct radiative forcing of the Earth's climate by aerosol particles or studies of ground-based or airborne atmospheric visual air quality.A seven-wavelength Aethalometer™ (Magee Scientific, model AE-31), described by Fialho et al. (2005), was used to measure the concentration of BC in units of ng m -3 at the wavelengths of 370, 470, 520, 590, 660, 880, and 990 nm at 5-min intervals from March 2007 to November 2008.We also used multiangle absorption photometer (MAAP) instruments to measure the concentration of BC at the wavelengths of 450, 500, and 700 nm from September 2009 to December 2010.Although we measured the concentration of BC with two different instruments, the outputs provided very comparable results (Hyvärinen et al., 2011).A comparison of values before and after a filter change showed that the aerosol light absorption is reduced by filter multiple scattering enhancement factors (Arnott et al., 2005).The light absorption coefficient (B ap ) at the given wavelength of 520 nm was calculated using B ap = m BC × 10 m 2 g -1 by assuming a specific absorption efficiency of BC of 10 m 2 g -1 , which is supported by Li et al. (2010) and Wang et al. (2015).In this study, the single scattering coefficient (SSA) was derived using the scattering coefficient and the absorption coefficients at the wavelength of 520 nm.Angstrom (1929) reported that the spectral dependence of extinction by particles could be expressed as a power law relationship: where τ(λ) is the aerosol optical thickness (AOT) at the wavelength λ and Å is the Ångstrom exponent.According to the relationship, the corresponding Å is calculated from the slope of the linear regression of the logarithm of the equation: Then, using the method of least squares, the scatter Ångstrom exponent (SAE) and absorption Ångstrom exponent (AAE) are obtained.The number concentrations of fine-mode (0.5 < D p ≤ 1 µm, hereafter N 0.5-1 ) and coarsemode (1 < D p ≤ 10 µm, N 1-10 ) particles were measured continuously using a 51-size channel aerodynamic particle sizer spectrometer (APS, Model 3321) with optical diameter in the range of 0.5-20 µm.Here, the size range was selected as 0.5 to 10 µm due to the detection limit of the APS and the large uncertainty of particle number concentration at sizes beyond 10 µm.

Temporal Variability of Aerosol Properties
Daily averaged aerosol scattering coefficients and aerosol absorption coefficients at the wavelength of 520 nm as well as PM 10 concentrations at SACOL from March 2007 to December 2010 are shown in Fig. 1; the statistical analyses are summarized in Table 1.Generally, the daily averaged B sp , B ap , and PM 10 were 155 ± 102 Mm -1 (hereafter, related results are shown by mean ± standard deviation), 16 ± 10 Mm -1 , and 120 ± 122 µg m -3 , respectively; these are much lower than the average values in Beijing, which are in the range of 488 ± 370 Mm -1 for B sp and 83 ± 40 Mm -1 for B ap , respectively (Bergin et al., 2001).We found high concentrations of PM 10 from November to February and from March to May, with average values of 121 ± 68 µg m -3 and 163 ± 199 µg m -3 , respectively.The variation in PM 10 was much larger in spring than in winter because of frequent dust events or local blowing soil dust during spring (Littmann, 1991;Wang et al., 2008Wang et al., , 2015)).The daily maximum PM 10 could exceed 2519 µg m -3 during dust events in spring.Previous studies have indicated that fossil fuel and biomass burning occur more dominantly from November to February (hereafter referred to as the cold season) over northern China, resulting in serious local air pollution (Wang et al., 2004;Sun et al., 2005;Li et al., 2007;Yuan et al., 2008).The maximum values of B sp and B ap were 894 Mm -1 and 85 Mm -1 , and the average values were 240 ± 145 Mm -1 and 26 ± 13 Mm -1 , respectively.Although the concentration of PM 10 was highest in March-  As a result, we conclude that the variations in B sp and B ap were not only determined by PM 10 but also depended on the particle size distribution and the compositions of chemical species (Hegg et al., 2009(Hegg et al., , 2010)).Statistically, the values of B sp , B ap , and PM 10 were much smaller in the period of June to October (hereafter, the warm season) because of more rainfall in this season.Fig. 2(a) shows the seasonal variations in the aerosol number concentrations of N 0.5-1 and N 1-10 (0.5 µm < D p ≤ 1 µm and 1 µm < D p ≤ 10 µm, respectively).The seasonal variations in N 0.5-1 and N 1-10 exhibited a seasonal cycle of higher concentrations in the cold season, with maximum values of 787 # cm -3 and 140 # cm -3 , compared with lower concentrations in the warm season, with minimum values of 3.9 # cm -3 and 0.8 # cm -3 , respectively.Both N 0.5-1 and N 1-10 increased remarkably in the cold season, and the number concentration of N 0.5-1 increased three times more than did that of N 1-10 .Obviously, the fine-mode particles were dominant in the cold season at SACOL, reflecting the role of local air pollutants originating from fossil fuel and biomass burning in the rural and remote regions.For the coarse-mode particles, N 1-10 was not only very high in the cold season but also remained at its highest number concentration in spring, which is characterized by local soil dust or dust events.Gong et al. (2003) noted that ~252.8Mt of soil dust was estimated to have been produced in the East Asian deserts in a spring season, with ~60% of this being due to four major dust storms.When dust storms occur over northern China, N 1-10 can increase sharply.These results showed good agreement with those of previous studies (Mori et al., 2003;Marinoni et al., 2008).In Fig. 2(b), the number concentration of dN/dlogD p showed a prominent seasonal cycle with a daily averaged maximum value of 4040 # cm -3 at 0.72 µm.During this field campaign, fine-mode particles with diameters less than 2 µm were dominant in this area, even during strong dust-event periods.The dN/dlogD p dropped dramatically along with the particle size distribution when the particle size was greater than 5 µm.Previous studies showed that the dN/dlogD p with particle diameters of 2-5 µm almost reached its highest values during dust events, suggesting that dust events make a great contribution to the increase in particle number concentration of coarse-mode particles (Bi et al., 2013;Huang et al., 2014).
The annual variations in BC concentration and N 0.5-1 are presented in Fig. 3.The boxes and whiskers denote the 10th, 25th, 50th (median), 75th, and 90th percentiles of the data, with dots marking average values.The BC concentration in this campaign was measured by AE-31 and MAAP from  The results indicate that the emissions of BC originated mostly from coal and biomass burning during the cold season.For comparison, the monthly median N 0.5-1 reached its highest value of 228 # cm -3 with larger variation in the cold season, and reached its lowest value of 17.5 # cm -3 in the warm season.Especially, there was a positive linear correlation between the daily average BC concentration and N 0.5-1 (R 2 = 0.52) at the 99% confidence level, implying very similar emission sources (Marinoni et al., 2008).The average values of BC and N 0.5-1 were larger (3-30%, 8-130%) than the median values, indicating that high air pollution was a large fraction of the contribution of BC and N 0.5-1 during the field campaign from March 2007 to December 2010.It was very interesting that the ratios of monthly averaged BC during June 2007 to February 2008 to the relative monthly averaged BC during July 2009 to November 2010 were 0.74 to 1.41 (1.08 ± 0.21: mean and standard deviation).However, the ratios of N 0.5-1 were in the range of 0.93 to 3.8 (average of 2.34 ± 0.77) and especially up to 1.76-3.8 in the cold season (average of 2.83 ± 0.82).The reason could be the limitation of industrial emissions to improve air quality.The number concentration of N 0.5-1 decreased significantly as a fraction of 36% in the cold season, and the potential reason could be concluded as the limitation of industrial emissions to improve air quality (An et al., 2007;Coatabile et al., 2010;Zheng et al., 2013).However, the concentration of BC, originating from biomass and fossil-fuel burning, was even higher in the cold season of 2009.

Seasonal Variations of Aerosol Properties
The averaged diurnal variations in B ap , B sp , SSA, and PM 10 over three periods are shown in Fig. 4. The minimum concentration of PM 10 was recorded during daytime at about 16:00 because of the strong vertical convection at that time.There were two peaks of concentration, whose timing varied in the different periods (Mar-May: 8:00 and 21:00, Jun-Oct: 8:00 and 19:00, Nov-Feb: 10:00 and 20:00) with average concentrations of 181.6 and 178.8, 104.0 and 93.8, and 149.0 and 125.5 µg m -3 , respectively.Generally, the concentration of PM 10 was higher in spring than in the other periods.Due to biomass and coal burning resulting in serious anthropogenic emissions, the concentration of PM 10 in winter was only lower than that in spring.The result agrees well with those of previous studies (Bi et al., 2010;Wang et al., 2010Wang et al., , 2015;;Che et al., 2014).As for PM 10 , the diurnal cycle of B ap presented two peaks (Mar-May: 8:00 and 21:00, Jun-Oct: 8:00 and around 20:00-23:00, Nov-Feb: 10:00 and 20:00), with average values of 15.3 and 11.2, 17.8 and 14.3, and 32.1 and 28.3 Mm -1 .Compared with the relatively low PM 10 concentration during the cold season, B ap was highest in the cold season, which can be explained easily because biomass and coal burning produced large amounts of light-absorbing aerosols, especially BC which has the ability to absorb visual light greater than that of the dust particles that were predominant in spring.The variation in B sp was consistent with that in B ap , with peaks of 248 and 290 Mm -1 , 156 and 129 Mm -1 , and 171 and 132 Mm - 1 in the cold season, spring, and warm season, respectively.The single-scattering albedo (SSA), defined as the ratio of scattering efficiency to total extinction efficiency, is a key parameter for investigating aerosol optical properties.During this field campaign, the SSA showed a trend opposite to the peaks of B ap and B sp .The SSA was lowest in the cold season due to more light-absorbing aerosols from biomass and coal burning, with two valley values of 0.87 at 10:00 and 20:00-23:00, and with a maximum value of 0.89.The SSA was largest in the warm season, with values ranging from 0.89 to 0.91, suggesting the lowest amount of lightabsorbing aerosols relative to light-scattering particles during this period.
The seasonal average size distributions of the fine and coarse modes were calculated from March 2007 to December 2010, as shown in Fig. 5.The highest averaged dN/dlogD p occurred with particle size distributions in the range of 0.5  to 1.84 µm in the cold season and 1.84 to 10 µm in spring.The peak values of dN/dlogD p were 198 # cm -3 at 0.72 µm in spring, 237 # cm -3 at 0.67 µm in the warm season, and 632 # cm -3 at 0.67 µm in the cold season.dN/dlogD p with diameter larger than 1.84 was a little higher in spring compared with the other two periods because of the influence of local blowing soil dust and frequent dust storms.As shown in Fig. 5, the trend of dN/dlogD p with particle size over the whole range was similar between the warm and cold seasons, reflecting the similar emission sources.The difference in the trends was that the aerosol emission in spring had considerably more large particles compared with the abundant fine-mode particles in the cold season, which was dominated by emissions from fossil-fuel and biomass burning for heating.Although the PM 10 concentration in the cold season was not the highest, the particle number concentration of light-absorbing aerosols was much higher than in the other periods.As a result, both B ap and B sp showed their highest values in the cold season.Comparing the spring and warm seasons, a pattern shift appeared at the size of 0.97 µm (near 1 µm), and the value of dN/dlogD p from 0.5 to 0.97 µm was a little higher in the warm season.The higher aerosol number concentration during the warm season may be related to the amount of anthropogenic emissions, such as industrial emissions and emissions from fuel burning, and to the fact that the nature of the emissions in spring, originating from local blowing soil dust and frequent dust storms, made a primary contribution to coarse-mode aerosols but had only a limited effect on fine-mode aerosols.Additionally, the result of light-absorbing particles, such as BC and OC, with sizes below 1 µm has already been discussed in previous studies (Petzold et al., 1997;Hitzenberger and Tohno., 2001).

Comparison of Aerosol Properties of Dust Events and Air Pollutants
Six cases each of dust events and anthropogenic air pollution events were selected to further examine their characteristics (Figs. 6 and 7).The statistical analyses of these events are summarized in Table 2.As shown in Fig. 6(a), the time series of PM 10 during the dust events were different and varied due to the meteorological conditions and the intensity of the dust emission sources.The dust events occurred at various times of the day, and their durations varied from several hours to dozens of hours.For example, the hourly averaged PM 10 at 14:00 (local time) on 29 February 2008 was around 100 µg m -3 , but the PM 10 then increased sharply, with large intensity but short duration, to its peak value of 2078 µg m -3 , nearly 20 times higher than  the value before the dust storm occurred.For another case occurring on 2 May 2008, two peak concentrations of PM 10 of 1945 µg m -3 and 1889 µg m -3 occurred at 3:00 and 7:00, respectively; then, the concentration of PM 10 decreased slowly to the minimum value of 434 µg m -3 at 19:00.At 2:00 on 3 May 2008, the PM 10 concentration increased to its peak value of 1519 µg m -3 after a few hours.However, these two nearly continuous dust events (2 and 3 May; intervals of only a few hours) originated from different emission sources according to the 60-h HYSPLIT air-mass back trajectories (Kalkstein et al., 1987;Stunder, 1997;Draxier and Hess, 1998;Draxler and Rolph, 2008;Huang et al., 2008a;Zhang et al., 2013).These air-mass back trajectories have been considered by previous studies to be an effective method of tracking the transport of dust events (Zhang et al., 2003a, b;Huang et al., 2010;Chen et al., 2013) and have been applied in deserts such as the Taklimakan (Li et al., 2007;Huang et al., 2008a;Wang et al., 2008;Zhou et al., 2013) and Mongolia Gobi (Xu 2004;Zhang et al., 2013).Here, we also previously indicated that local blowing dust could contribute to very high PM 10 concentration.Additionally, the variations in B ap and B sp were generally consistent with the changes in PM 10 .As an example, the values of B ap and B sp decreased from maximum values of 23.3 and 143 Mm -1 to minimum values of 11.3 and 50 Mm -1 with maximum/ minimum ratios of around 2 and 3, respectively, during the morning of 10 April 2008, but the decreasing trends were smaller than that of the PM 10 variation.While the dust event occurred, the increased SSA was consistent with the similar changes in PM 10 .Similar results were obtained previously by Wang et al. (2015).Similar to Fig. 6, diurnal cycles of PM 10 , B ap , B sp , and SSA are shown during local pollution episodes in Fig. 7. Comparing their optical properties, the pollution episodes could exist for a long period of time according to the local weather conditions.For instance, on 9 January 2008, the averaged PM 10 concentration was 228 ± 50 µg m -3 , with a range of 108 to 301 µg m -3 and a valley that appeared at about 16:00.In contrast to the dust events, the value of B ap was very sensitive to the variation in PM 10 because of the large amount of light-absorbing impurities in the cold season, as we discussed in the previous section.On 6 January 2008, the PM 10 concentration and B ap decreased from 407 to 44 µg m -3 and from 90.9 to 9.4 Mm -1 , respectively.Additionally, the variation in B sp showed the same trend as those in PM 10 and B ap.The variation in SSA had no significant trend corresponding to those of PM 10 , B ap , or B sp .For instance, the variation trend of SSA was opposite that of PM 10 on 6 January 2008, but it had a similar trend on 3 February 2008.
The lowest value of SSA was 0.71 ± 0.03 on 9 January 2008, but SSA was almost always above 0.8 in the other cases.
Dust events and anthropogenic air pollution were two important weather conditions affecting local air quality at SACOL.To acquire deeper knowledge of the differences of aerosol size distribution and optical properties between dust events and anthropogenic pollution, a comparison was conducted (Fig. 8).The comparison of B ap and PM 10 is presented as six cases of pollution episodes (red dots) and dust events (black dots).The comparison shows a clear distinction of lower PM 10 with higher B ap for pollution episodes and higher PM 10 with lower B ap for dust events.The best fit of the slope could reflect the PM 10 mass absorption efficiency and mass scattering efficiency (Ayers, 2001;Hammer et al., 2001;Garland et al., 2008).The slope of B ap /PM 10 was 0.132 for pollution episodes, much higher than the 0.008 for dust events, showing the markedly higher mass absorption efficiency of aerosols such as BC and OC in anthropogenic pollution.The correlation coefficient between B ap and PM 10 was R 2 = 0.53, suggesting more similar sources.As Fig. 6 shows, a dust event passed through SACOL from 16:00 to 23:00 on 29 February 2008.In the morning, the B ap was around 15 Mm -1 , with a maximum of 20.7 Mm - 1 before the dust storm occurred, but B ap then decreased to a minimum of 3.5 Mm -1 followed by an increase of PM 10 during the dust event.The B ap then slowly increased to its peak (16.7 Mm -1 ), which was close to its values in the morning.We note that the B ap was dominated by mineral dust during the dust event of 29 February 2008 but that it was affected by the combined effects of dust storms and local pollution in other cases.The results indicate that mineral dust was dominant during a heavy dust event.Additionally, the mass absorption efficiency of dust and BC (mainly from local pollutants) is highly different (Zhang et al., 2007;Yang et al., 2009).As a result, mineral dust mixing with local pollutants led to a lower correlation coefficient between B ap and PM 10 .Similar to the case for B ap , the slope of B sp /PM 10 was 0.61 during pollution episodes and 0.12 during dust events, reflecting the higher mass scattering efficiency of pollution events.The correlation coefficient between B sp and PM 10 of R 2 = 0.21 was lower than that for dust events (0,64), reflecting more complicated compositions of aerosols during the pollution episodes.A scatter plot of scattering aerosol efficiency (SAE) and absorption aerosol efficiency (AAE) is shown in Fig. 8(c).For dust events, the ranges of AAE and SAE were 1.25 to 1.75 and 1.0 to 2.5, respectively.Comparably, the ranges of AAE and SAE were mainly from 1.0 to 1.5 and 1.0 to 4.0, respectively, during pollution episodes.On 29 February 2008 and 8 May 2008, the PM 10 was around 64 µg m -3 at13:00 and 119 µg m -3 , respectively, at 14:00, before the dust event occurred.The PM 10 concentration and AAE were in the ranges 1070-2080 µg m -3 and 1.55-1.74(17:00-23:00) and 240-830 µg m -3 and 1.24-1.28(15:00-19:00), respectively, during the dust events.The mixing ratios of local aerosols (including mainly local air pollutants) to aerosols during the dust events were 0.03-0.06and 0.14-0.5.The results indicated that the AAE of mineral dust was beyond 1.5, which was similar to the results of Dubovik et al. (2002) andRussel et al. (2010).The values of AAE were much lower during air pollution than during dust events.However, there were slight changes in SAE due to the amount of mineral dust.As shown in Fig. 8(d), the peak value of dN/dlogD p for pollution events was up to 1500 # cm -3 at 0.7 µm.During dust events, the peak value was below 200 # cm -3 at around 1.0 µm (Sow et al., 2009).The variation in dN/dlogD p was very small at diameters of 0.5-2.0µm, as indicated by Sow et al. (2009).Large differences in dN/dlogD p between dust and air pollution events occurred at around 1.0-2.0µm, and dN/dlogD p was ~10 times higher for fine-mode particles during air pollution events than during dust events.

CONCLUSIONS
In the field experiment, the average values of B sp , B ap , and PM 10 concentration from March 2007 to December 2010 were 155 ± 102 Mm -1 , 16 ± 10 Mm -1 , and 120 ± 122 µg m -3 , respectively.The seasonal changes in PM 10 were higher, with values of 121 ± 68 µg m -3 and 163 ± 199 µg m -3 in the cold season and spring season, respectively.The variations in B sp and B ap were much smaller compared with the changes in PM 10 .The high values of B sp and B ap appeared in the cold season, with average values of 240 ± 145 Mm -1 and 26 ± 13 Mm -1 , respectively.The particle number concentration showed that both fine-mode and coarse-mode particles increased significantly in the cold season; however, the number concentration of fine-mode particles increased much more than that of coarse-mode particles.
The dust events occurred at various times of the day and lasted from several to dozens of hours.In contrast, the pollution episodes could exist for a long period of time, depending on the local weather conditions, and the B ap was very sensitive to changes in PM 10 during the air pollution events.The slopes of B ap /PM 10 and B sp /PM 10 could reflect the absorption and scattering properties of the aerosols.Generally, the slope of B ap /PM 10 was 0.132 for pollution episodes, which was much higher than the 0.008 value for dust events, showing markedly higher mass absorption efficiency for light-absorbing aerosols such as BC and OC originating from anthropogenic pollution.The correlation coefficient between B sp and PM 10 was higher during the dust (R 2 = 0.64) than during the air pollution events (R 2 = 0.21), reflecting the more complex compositions of the aerosols during the air pollution events during the cold season.In general, the dominant sources of PM 10 , BC, and the other aerosol types in this region were obviously complex.We note that further field experiments for measuring dust events interacting with local pollutants near the dust sources are needed.

Fig. 1 .
Fig. 1.Variations in daily averages of (a) aerosol absorption coefficient (B ap ) at 520 nm, (b) aerosol scattering coefficient (B sp ) at 520 nm, and (c) PM 10 concentration at SACOL from March 2007 to December 2010.

Fig. 3 .
Fig. 3. Seasonal variations in BC concentration and the number concentration of fine-mode particles (N 0.5-1 ) at SACOL from March 2007 to December 2010.

Fig. 6 .
Fig. 6.Hourly average variations in (a) PM 10 concentration, (b) B ap , (c) B sp , and (d) SSA at SACOL during dust events from March 2007 to December 2010.

Fig. 7 .
Fig. 7. Hourly average variations in (a) PM 10 concentration, (b) B ap , (c) B sp , and (d) SSA at SACOL during local air pollution events from March 2007 to December 2010.

Fig. 8 .
Fig. 8.Comparison of hourly aerosol optical properties and daily number concentration between dust events (black dots) and local air pollution events (red dots) at SACOL from March 2007 to December 2010.

Table 1 .
Statistics based on measured aerosol optical properties obtained during the field experiment.
ap (Mm -1 ) 16 ± 10 11 ± 5 14 ± 7 26 ± 13 SSA 0.90 ± 0.06 0.89 ± 0.07 0.91 ± 0.04 0.88 ± 0.07 May of all seasons, the values of B sp and B ap did not significantly increase and were only about half those in the cold season.Li et al. (2010) indicated similar results at the National Climate Observatory of the China Meteorological Administration (39.082°N, 100.276°E; 1,460 m a.s.l.), which is located ~20 km northwest of Zhangye, Gansu Province.

Table 2 .
Statistics of aerosol optical properties observed during local air pollution and dust events.