Airborne Observations of Cloud Condensation Nuclei Spectra and Aerosols over East Inner Mongolia

Airborne particles with the capability of accelerating the condensation of water vapor into cloud droplets under specific conditions are defined as cloud condensation nuclei (CCN). The concentrations of CCN within the lower troposphere have complex effects on microphysical processes within clouds and, furthermore, on many aspects of climate and weather (Andreae and Rosenfeld, 2008). These interactions have been documented in many studies addressing the effects of aerosols on cloud processes and climate change (e.g. McFiggans et al., 2006). High CCN concentrations (N _CCN) tend to increase the number concentration of cloud droplets, reduce their size and collection efficiencies and, consequently, decrease precipitation efficiency, which in turn modifies cloud microphysics. This process results in increased planetary albedo (Towmey, 1977), cloud lifetime (Albrecht, 1989), and spatial extent in sequence (aerosol indirect effect; Hudson and Yum, 2001). Previous observations and numerical modeling further support the finding that aerosols from polluted air masses decrease the cloud droplet radius and suppress rainfall (Nober et al., 2003). However, there are still large uncertainties about the potential effects of aerosols on clouds, as well as climate change, due to limited in-situ observations of vertical distributions of key variables, the inadequacy of existing tools, and a failure to account for small-scale processes that buffer cloud and precipitation responses to aerosol perturbations (Stevens and Feingold, 2009). Most climate and detailed microphysical models are at present in great need for information on the vertical distribution of aerosols and CCN number concentrations. Therefore, to incorporate the potential influences of CCN in meteorological models at all scales, knowledge of the temporal and spatial distributions of CCN in the lower troposphere is very important. For decades, measurements of CCN spectra over wide areas of Earth's surface have been collected at a variety of marine and continental sites (e.g. Hobbs, 1971; Hoppel et al., 1973; Hudson and Xie, 1999; Roberts et al., 2001; Vestin et al., 2007; Paramonov et al., 2015). These studies suggested that the shape of CCN spectra reflects the influence of particle size distributions and the size-dependent chemical composition. In addition, aircraft observations of aerosols and clouds, considered the most direct and effective data source, are capable of obtaining high-resolution in-situ data of aerosol and CCN vertical distributions directly observed in the lower troposphere (Haywood et al., 2003; Zhang et al., 2006).

In the last 20 years, airborne data collected in several provinces of North China have showed that Beijing and the surrounding areas have become one of the most polluted areas in China. Several studies have demonstrated that the aerosol and CCN distributions in this region are strongly affected by the atmospheric boundary layer and long-range transport of pollutants. Increasing anthropogenic emissions of aerosol particles and biomass burning tend to impact regional air quality, precipitation and climate (e.g. Zhang et al., 2006; Shi and Duan, 2007, Deng et al., 2009, Liu et al., 2009; Rose et al., 2010; Duan et al., 2012; Lu and Guo, 2012). A recent study indicated a significant decrease in the magnitude of precipitation in eastern Central China and a correlation between the reduction of precipitation and high concentrations of aerosols during the last 40 years (Zhao et al., 2006). Note that aerosol emissions from megacities in China not only affect the local environment and cloud microphysics, but also the downwind aerosol chemical composition and cloud microphysical characteristics via long-range transport under specific atmospheric circulation conditions. Several in-situ observations over Korea, Japan and the Yellow Sea have been conducted to investigate the CCN and the hygroscopic properties of aerosols from the East Asian continent (Jaffe et al., 2003; Adhikari et al., 2005; Kim et al., 2011). However, the limited spatial and temporal coverage makes it difficult to quantify the effects induced by aerosols in this region. Thus, the potential effects of CCN on climate change and regional precipitation are still unclear.

The focus of this study is to present a summary of measurement results of aircraft observations over the Tongliao area, East Inner Mongolia, China. Attempts are made to characterize the monthly variations in CCN spectra and aerosol number concentrations within the boundary layer downwind of Beijing and its surrounding areas. This study also includes representative values of vertical distributions of CCN spectra and aerosol concentrations under relatively clean and polluted conditions. It aims to document the characteristics of CCN and aerosols during flights and record the impacts of pollutants on CCN spectra and activation efficiencies. The results can be used in further studies, including aerosol indirect effects, as well as initializations for detailed cloud resolving microphysical models (Yin et al., 2000; Chen et al., 2011) of precipitation processes in this region.

Located northeast of Beijing, Tongliao is a light industrial town with its rural area mainly covered by sands and alluvial plains. Changchun and Shenyang are two large cities lying to the east and southeast (Fig. 1) of Tongliao City, respectively. To investigate cloud systems passing through the study area and aerosol-cloud interactions in an area downwind of a heavily polluted megacity, several in-situ observations were conducted from 2009 to 2015. They were performed by the Laboratory of Cloud-Precipitation Physics and Severe Storms, Institute of Atmospheric Physics, associated with the Tongliao Meteorology Bureau under the support of the Strategic Priority Research Program of the Chinese Academy of Sciences. During the field study, cloud, aerosol and CCN were measured using a set of state-of-the-art airborne instruments. To explore the vertical distributions of aerosols and CCN, the flight strategies used in this paper included the combination of many constant altitudes runs (300 m each level) with profiled descents and simple descents. The data shown in this paper were mainly obtained during two periods: from 14 May 2010 to 30 September 2010, and from 18 May 2011 to 30 August 2011. Throughout the in-situ observations, three to five flight soundings were performed every month, mostly in the region located 50-100 km north and northwest of Tongliao City and in cloud-free air to minimize the effects of cloud droplets and the city plume. The vertical distributions of CCN spectra presented in this study were obtained through soundings over the Tongliao region. The dataset includes standard state variables, winds and detailed microphysical parameters.

Figure 1.  Map showing locations mentioned in the main text. The study area is represented by a black rectangle. The double circle and single circle stand for Beijing and Tianjin City, respectively. The red and blue dots are Shenyang and Changchun cities, respectively.

Instruments employed in the observations were mounted on a twin-engine aircraft (Yun-12) to simultaneously obtain the size spectra of aerosol particles, CCN number concentrations at different supersaturations, meteorological parameters and spatial locations of the aircraft. The instruments utilized in this study are summarized in Table 1. The size distributions of airborne particles were obtained using a particle measuring system (DMT Inc., Boulder, CO, USA), involving a cloud and aerosol spectrometer (CAS) and a passive cavity aerosol spectrometer probe (PCASP). Particles with diameters ranging from 0.1 to 3.0 μm were obtained by the PCASP with 30 unequally sized bins, while larger particles were measured by the CAS with a sampling range from 0.6 to 50 μm. 1-Hz data, including aerosol and cloud droplet concentrations, temperature, relative humidity and other meteorological parameters were provided for analysis. Note that the aerosol particles concentrations (N _a) were obtained from the PCASP in this paper, while the in-cloud data were excluded to avoid the effects of cloud droplets on measurements. Additionally, the sizes of particles in clear air partly depend on relative humidity; thus, the vertical variation in N _a will be due to humidity differences at different levels, resulting in N _a not always reflecting the actual differences in the aerosol. Therefore, to eliminate the effect of relative humidity on particle spectra, the data obtained under high humidity conditions (>70%) were also not considered. Meanwhile, the aerosol data obtained within the cloud were also excluded, since they were not reliable. Other meteorological parameters and the flight track were provided by the air data probe (AIMMS20) with a GPS module.

The total number concentration of aerosols is the sum of number concentrations from the lowest size bin to the highest one of the PCASP, which can be expressed by \begin{equation} \label{eq1} N_{\rm tot}=\sum_i{N_i} , \ \ (1)\end{equation} while the effective radius is derived using the equation \begin{equation} \label{eq2} R_{\rm e}=\frac{1}{2}\frac{\sum\limits_i{N_i D_i^3}}{\sum\limits_i{N_iD_i^2}} , \ \ (2)\end{equation} where D_i is the geometric mean diameter at the ith bin, N_i represents the number concentration of the ith bin, and R _e indicates the effective radius.

The aerosol size distribution is calculated by the equation \begin{equation} \label{eq3} f(\log D_i)=\frac{N_i}{\Delta\log D_i} , \ \ (3)\end{equation} where \(\Delta\log D_i=\log D_i,\rm upper-\log D_i,\rm lower\), and D_{i, upper} and D_{i, lower} represent the upper and lower boundary of the ith bin, respectively.

CCN concentrations were obtained using a stream-wise cloud condensation nuclei counter (CCNC, Droplet Measurement Technologies) (Roberts and Nenes, 2005; Lance et al., 2006) at a set of sequential supersaturations (from 0.2% to 1.0% at 0.2% intervals) and with a 1-Hz sampling resolution. Each CCN sampling cycle took 25 min (5 min for each supersaturation). The supersaturation in the CCNC was controlled by modifying the temperature gradient between the bottom and top of the inlet, column pressure, sample and sheath flow rates, as well as the temperature within the column. It should be emphasized that the supersaturations within this instrument are highly sensitive to pressure variations during flight sampling due to the limited time needed for the formation of the supersaturation and temperature profile in the column. To avoid such a shortcoming, the pressure at the inlet of the CCNC was maintained at ∼500 hPa by means of a DMT inlet pressure controller, which is a vacuum pump with the capability of controlling downstream of the flow restrictor. Additionally, the data obtained by the CCNC during first 30 s of each supersaturation were not used, because of the adjustment of temperatures and supersaturations within the column.

To verify the performance of the instruments, every year, before the in-situ observations, the CCNC was calibrated at various pressures, flow rates and temperature gradients by using ammonium sulfate particles generated by an atomizing dissolved mobility analyzer. The PCASP was also calibrated, using polystyrene latex spheres.

Additionally, the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler, 1992) was used to calculate the back trajectories of air masses and to identify the aerosol history of the long-range transport. Because aerosol characterization depended on the simulation results of the HYSPLIT model in this paper, data from the NCEP's Global Data Assimilation System at a temporal and spatial resolution of 3 h and 1^°× 1^°, respectively, were used for the initial meteorological data, which is necessary when starting a back-trajectory simulation. Meanwhile, the starting location for each trajectory was based upon the actual location of the aircraft sounding.

Figure 2.  Vertical profiles of the aerosol number concentration in this and previous studies. The solid green lines, red line and black short-dashed line stand for the vertical profiles of N _a in each case, the mean aerosol number concentration (N _a-ave), and the regression function of N _a-ave in this study, respectively. The dot-dashed blue line is the regression function of the mean N _a profile obtained over the Beijing region, which was obtained in February 2005 to September 2006 using the PCASP (0.16-3 μm; Liu et al., 2009). The dashed blue line represents the regression function of the mean N _a over the clean continent, provided by (Seinfeld and Pandis, 1997).

An intercomparison of the vertical distributions of aerosol number concentrations from Tongliao, the Beijing region and the clean continent was retrieved from in-situ observations (Fig. 2). The vertical profiles of aerosol concentrations in Beijing (polluted area) presented in this study were obtained from February 2005 and September 2006 using the PCASP (Liu et al., 2009), while the counterparts measured over the typical clean continent were from (Seinfeld and Pandis, 1997). The results showed that the aerosol number concentrations (N _a) measured by the PCASP near the surface level ranged from 200 to 4000 cm^-3 in the Tongliao region. The average value (approximately 2000 cm^-3) was much smaller than that of Beijing but similar to that of the typical clean continental profile. The average value of N _a decreased with an increase in altitude (red line in Fig. 2), which can be expressed by the classical exponential equation \begin{equation} \label{eq4} N_{\rm a}=N_0\exp\left(-\frac{H}{H_{\rm p}}\right) , \ \ (4)\end{equation} where H _p (1733 m) and N₀ (1785 cm^-3) represent the scale height and aerosol number concentration near the surface, respectively. This exponentially declining profile of aerosol number concentration is similar to that of many previous studies (e.g. Seinfeld and Pandis, 1997). Note that the N₀ is also lower than that measured in most places of the North China Plain. By analyzing airborne observations measured in North China, (Duan et al., 2012) proposed that a value of 4118 cm^-3 is the typical N₀ for the vertical distribution of aerosol in the North China Plain. As mentioned above, the average aerosol concentration profile indicated that aerosol particles from the surface were much smaller than those of the Beijing region and North China Plain, but similar to the clean continental types. However, N _a profiles vary from case to case, possibly due to the influences of monthly variation and the long-range transportation of pollutants under specific atmospheric circulation. These potential effects on aerosol distributions in East Asia have been well documented by many researchers (e.g. Adhikari et al., 2005; Kim et al., 2011). In this paper, we mainly focus on the monthly variations of aerosol and CCN, as well as the influence of pollutants from North China on the downwind area. The following sections analyze these cases in detail.

Figure 3.  Average distributions of MODIS AOD (1^°× 1^° resolution; shaded area) and wind field (NCEP reanalysis data; arrows): (a) May-July 2010; (b) August-September 2010; (c) May-July 2011; (d) August-September 2011. The black rectangle shows the location of the study area (Tongliao), and the double circles indicate Beijing City. White shading represents missing data.

To confirm the reliability of our measurements, a rough comparison with Moderate Resolution Imaging Spectroradiometer (MODIS) data was made. MODIS-derived aerosol optical depth (AOD) is generally related to dust and anthropogenic emissions, high relative humidity, precipitation, and regional atmospheric dispersion (Kim et al., 2007). Figure 3 depicts the distribution of averaged MODIS-derived AOD at 550 nm wavelengths in North China and the flight area in this study. The average MODIS AOD distribution was derived from NASA level 2 daily products. Data from May to September 2010 and May to September 2011 were averaged in 1^°× 1^° grids. A comparison of the MODIS AOD in the flight area around the Beijing area and our study further proves that Tongliao City was relatively clean (Fig. 3). The average MODIS AOD value of the flight area (black rectangle in Fig. 3) was approximately 0.4 from May to July and smaller than 0.2 in August to September. In contrast, in the heavily polluted region stretching from Beijing, Southeast Tianjin and western Shandong Province to the northern and eastern areas, the average AOD exceeded 0.9 in both May to July and August to September. Meanwhile, average MODIS AOD also demonstrated that the north, east and west areas to Tongliao City were much cleaner than Beijing and its surrounding area. The variations of AOD from May to September in those areas had the same trend as Tongliao City. Moreover, an increase in AOD in the Tongliao area in May-June was significantly modulated by the transport pattern and strength of the southeasterly outflow of the heavily polluted area.

Figure 4.  HYSPLIT (72-h) back trajectories within the boundary layer for each CCN measurement (at 1000 m). Red lines represent air masses from polluted areas, and blue lines represent relatively clean air masses from remote continental areas.

All aerosol number concentrations used in this study showed similar trends to the MODIS-derived AOD. Table 2 shows the averages and standard deviations of N _a (accumulation mode aerosol: 0.1-3 μm) and N _CCN within the boundary layer (BL; altitude less than 1500 m) over the study area. The average N _a within the BL in May to July was greater than that in August to September. The maximum monthly aerosol number concentration occurred in May 2010 and May 2011 (2455 674 cm^-3, 1868 831 cm^-3), and the minimum in September 2010 and August 2011 (below 1000 cm^-3). Monthly variations of aerosols over East Asia have been analyzed in many previous studies. For example, by employing multi-year records of micro pulse lidar data, MODIS data and 850-hPa wind vector NCEP reanalysis data, (Kim et al., 2007) revealed that the aerosol maximum occurred in May to June, as a result of an increase in anthropogenic aerosols and a stagnant synoptic meteorological system overlying the East Asian continent. On the contrary, from July to September, the East Asian continent was mainly influenced by wet air masses from the ocean. Thus, a large amount of aerosols was removed by frequent precipitation. In September, the dispersion of aerosol loadings was strongly related to fast-moving synoptic weather patterns, such as the prevailing northwesterly wind at 850 hPa. However, no obvious trend of seasonal or monthly variation of aerosol concentrations was found over the heavily polluted areas in North China (Shi and Duan, 2007; Duan et al., 2012). Hence, we speculate that aerosol number concentrations during the aircraft soundings were affected by many complex mechanisms, including wet scavenging, city plume bursts and sand storms.

Monthly variations of N _CCN over the Tongliao area are presented in Table 2. The similarity between the monthly variations of N _CCN and N _a suggests that CCN concentrations in the BL were strongly related to aerosol particle concentrations. Additionally, the aerosol activation efficiency, defined as the ratio of N _CCN to N _a, representing the percentage of the aerosol that can be activated as CCN at a certain supersaturation, varied monthly. Generally, CCN activate less efficiently in May to June than July to September at all supersaturations, possibly because of the influence of organic aerosol particles, as well as rainout of soluble material in the wet season.

As mentioned above, the average MODIS retrieval data showed a distinct pattern in the East China area (Beijing and its surrounding area), with high AOD that ranged from 0.5 to 1.0. Therefore, in this study, to investigate the effects of aerosol on CCN spectra, the data were roughly divided into two categories based on the 72-h HYSPLIT back trajectories at 1000 m for each case (Fig. 4), as well as the average MODIS AOD. Under clean conditions, the air masses mainly came from the remote continental areas of Mongolia and Siberia (Russia), where the average AOD was usually less than 0.5 in May to September. Under polluted conditions, the air masses originating from heavily polluted industrial urban areas (e.g. Beijing City, Shandong Province and Hebei Province), had an average AOD usually exceeding 0.5, and contained aerosols dominated by fine particles (Kim et al., 2007) and possibly a large fraction of sulfates, black carbon, nitrates and organic matter, because of the increase in the burning of fossil fuels (Huebert et al., 2003, Kim et al., 2005). Both conditions exhibited distinctly different CCN concentrations at varying supersaturations (Fig. 5) within the BL. On average, CCN concentrations at all five supersaturations under polluted conditions were higher than those under clean conditions.

Due to difficulties in measuring the aerosol size composition distribution or its calculation from CCN properties, a popular and simple empirical parameterization was adopted to represent the relationship between CCN concentrations and supersaturations, i.e. the empirical power law given by (Twomey, 1971): \begin{equation} N_{\rm CCN}=cS^{k} , \ \ (5)\end{equation} where N _CCN is the total number concentration of CCN, c is a constant for the number concentration of CCN particles at 1% supersaturation, and k is the slope index. The fitting parameters are shown in Fig. 5. Our measurements show that parameter c under clean conditions (539 cm^-3) was lower than that under polluted conditions (1615 cm^-3); and under both conditions, the slope k indexes were similar (1.47 under clean conditions and 1.43 under polluted conditions). Compared with the CCN spectra in the boundary layer in the polluted or clean continental air masses of other parts of the world, the average CCN concentration under polluted conditions at higher S values (1%) was higher than that in typical marine cases (Hegg et al., 1991) but lower than that measured in Southeast China (Fang et al., 2016) and India (Konwar et al., 2012) (Table 3).

It takes significant time to measure the CCN spectra at different altitudes (25 min for each supersaturation cycle) due to the effects of air traffic control, aircraft maximum cruising time, and clouds/precipitation; thus, only two complete clear-air soundings of CCN spectra at different altitudes under cloudless conditions were obtained during the two years. This section examines the measured aerosols and CCN in these two cases by analyzing the total number concentrations (cm^-3), the effective radius of particles, and CCN spectra. The aerosol spectra were averaged over various periods corresponding to the sampling period of the CCN counter. The two flights were designed to follow basically the same pattern: a sampling location was selected, and the aircraft spiraled down to 600 m (above sea level) with a 300 m interval in clear air (Fig. 6).

Figure 5.  CCN concentrations at different supersaturations (referred to as CCN spectra) within the boundary layer (doted lines); all data obtained in 2010 and 2011 were used. The vertical bars stand for the standard deviation. The solid line represents the fitted function: (a) clean and (b) polluted.

Figure 6.  Flight track of aircraft on (a) 8 August 2010 and (b) 15 August 2010.

Figure 7.  Backward trajectories simulated by the HYSPLIT model over 72 h: (a) 2010 August 8 (polluted) and (b) 2010 August 15 (clean).

To obtain the source of the polluted air on 8 August, back trajectories were calculated at the levels of 500 m, 900 m and 2600 m over the observation area using the HYSPLIT model. The 72-h back trajectories indicated that the air passed through heavily polluted areas, such as Beijing, Tianjin and the south of Tongliao (anthropogenic sources), and arrived at the observation area on 8 August 2010 at 500-900 m (Fig. 7). However, at higher levels, the trajectories show that the air passed through the south of Mongolia. On the relatively clean day (15 August 2010), the transport paths indicated that the aerosol particles mainly originated from eastern Mongolia and northern Tongliao, which are relatively clear regions.

The flight was conducted during a period of heavy pollution, with a shallow trough located to the west of Tongliao City. Measurements of CCN spectra were made west of Tongliao; a clear-air sounding was obtained from 2300 UTC 7 August to 0140 UTC 8 August. Figure 8a shows the profiles of relative humidity, temperature and the number concentration of aerosol particles that were obtained from the PCASP. Following (Liu et al., 2009), the sounding data were averaged at 100-m height intervals and statistical outliers were eliminated.

Figure 8.  Vertical distributions of aerosols, CCN spectra, temperature and relative humidity on 8 August 2010: (a) relative humidity and temperature; (b) aerosol number concentration and effective radius; (c) size distributions of aerosol at different levels; (d) CCN spectra at different levels.

On that day, relatively polluted conditions were observed, with a large fraction of aerosol particles located between the surface and the base of the BL inversion (approximately 1000 m), as shown in Fig. 8b. Maximum aerosol concentrations in excess of 2500 cm^-3 were observed in this region; whereas, above the BL, values decreased to 1200 cm^-3 and the vertical variation was not obvious from 1200 to 2500 m. From an altitude of 2500 to 4500 m, the aerosol concentration further decreased to approximately 80 cm^-3. The CCN concentrations measured at different supersaturations (Fig. 8d) showed a distinct pattern of variation. At all levels, the CCN spectra were consistent with an empirical curve given by Eq. (5). The fitting parameters (c and k) and the determined coefficients are shown in Table 4. No definite trend was observed in the vertical distributions of CCN spectra. Spectra at all levels were steeper than the usual curve, N=cS^k, for most continental air (Table 3). The aerosol number concentrations at higher levels and the coefficients k (4500 and 3000 m) were much smaller than those at lower levels (2100 and 1000 m), which indicates different origins for CCN at different levels.

Figures 8b and c illustrate the vertical distribution of R _e and average particle size distribution (PSD) at different altitudes. With very small variability and an average value of 0.12 μm, most R _e values on that day ranged from 0.10 to 0.16 μm. Figure 8c depicts the vertical profile of size distribution. Above the boundary layer, the size distributions show clear similarity in shape, and the concentrations at all sizes decreased with height.

Figure 9.  As in Fig. 5 but for 15 August 2010.

During this flight, the aircraft observation was conducted from 0329 UTC to 0559 UTC in the unstable region (west of Tongliao City) behind a northeast-southwest cold front on 15 August 2010. The CCN and aerosol particles are shown in Fig. 9. Fairly clean conditions and northwesterly winds at all levels were observed. A clear-air sounding was obtained during the flight. The aerosol concentration was approximately 600 cm^-3 in the BL up to the inversion (approximately 1700 m; Fig. 9a). A more complicated distribution was observed above the BL, likely resulting from a different horizontal advection. The vertical profile of aerosol size spectra is presented in Fig. 9c. Similar spectra with peaks ranging from 0.15 to 0.18 μm at all levels revealed a sufficient mixing of aerosols.

The R _e vertical profile (Fig. 9b) shows completely different features compared with 8 August. The average R _e from the near-surface level to the top of the BL ranged from 0.30 to 0.45 μm, much greater than on 8 August (Fig. 9b). However, it decreased to 0.15 μm in the free troposphere (above 1500 m), and further to 0.1 μm at 4500 m. The PSDs at different levels (Fig. 9c) reveal a decrease in concentrations at all sizes with an increase in height and a reduction in the number concentration of fine aerosol particles (0.1<d<0.6 μm), as compared with that on 8 August at the lower levels. Therefore, fine particles, which may consist of sulfate and organic carbon (Gunthe et al., 2011), contributed little in this case.

In accordance with the CCN spectra presented in Fig. 9d, the maximum concentration at all supersaturations appeared at 800 m and decreased as the height increased. A smaller concentration of aerosols resulted in a smaller coefficient c compared with that on the polluted day. Meanwhile, the coefficients (k) of most empirical functions were greater than those on 8 August at the same levels. It should also be emphasized that the CCN spectra measured on 8 and 15 August at all levels were very similar (apart from 8 August at 4000 m), indicating a similar chemical composition or mixing state at different levels. However, this homogeneity is also relevant to the possible mixing of CCN at different levels, due to the fast displacement of an aircraft. For example, each supersaturation cycle took 25 min in our study and, during this significant time, the aircraft flew through different heights and the CCNC may have encountered different air masses, which in turn would have resulted in the artificial mixing of CCN at different levels. Therefore, it was hard to accurately distinguish between CCN spectra at specific levels in this study. According to the scanning flow CCN analysis approach proposed by (Moore and Nenes, 2009), this problem could be addressed properly in future studies by modifying the flow rate in the growth chamber over time, while maintaining a constant temperature gradient.

The impact of air pollution on the increase in CCN has been the subject of conflicting results and conclusions. Generally, inorganic particles are larger than organic particles (Broekhuizen et al., 2006; McFiggans et al., 2006). Moreover, inorganic particles are generally more CCN-active than organic particles (Raymond and Pandis, 2003). CCN concentrations obtained during the First Aerosol Characterization Experiment at Cape Grim, Tasmania, revealed that aerosols with diameters between 0.08 and 0.2 μm, Aitken mode particles (0.02 μm <d<0.07) and larger sea-salt (d>0.2 μm) accounted for 71%, 13% and 16%, respectively (at 0.5% supersaturation). When the aerosols were affected by local biomass burning or anthropogenic sources, the relative contributions were 80%, 6% and 14%, respectively (Covert et al., 1998). Furthermore, particles that contain different chemical compositions have different sizes. Thus, the nucleation capability of aerosol particles is generally related to their chemical composition and sizes.

The relationship between CCN and aerosols reflected the fact that CCN concentrations increased with an increase in the aerosol concentration under polluted conditions. Despite substantially high CCN concentrations under heavily polluted conditions (8 August 2010), the slope of N _CCN/N _a was 0.74 (0.4% supersaturation); whereas, under the relatively clean conditions, it was 0.87 (Fig. 10). To compare the N _CCN/N _a in the Tongliao area with that of previous research, data from other in-situ observations are summarized in Table 5. (Raga and Jonas, 1995) suggested a logarithmic relationship between CCN and aerosol particles, with a log slope of 0.44. (Hegg et al., 1996) showed that only 10% of the particles with diameter >3 nm can be identified as CCN at 1% supersaturation. (Chuang et al., 2000) obtained a log slope of 0.63 for aerosol particles with diameters >0.1 μm and CCN at 0.1% supersaturation. (Adhikari et al., 2005) found the concentration of CCN was strongly related to the concentration of aerosol particles, according to the linearly relation N _CCN≈ 0.75N _a.

The slope of N _CCN/N _a was lower under polluted conditions than that under clean conditions in the Tongliao area. Likewise, (Raga and Jonas, 1995) uncovered a similar finding in that the CCN concentration, usually higher than the aerosol concentration on clean days (at a 0.85%-0.95% supersaturation rate), was lower than the aerosol concentration on one of the polluted days around the British Isles. The observation was attributed to the presence of a large fraction of carbon from gas/oil rigs. (Lu et al., 2008) and (Leaitch et al., 1986) similarly revealed that, in continental cumulus cloud, the droplet activation efficiencies were high for low N _a and low for high N _a. Thus, it may be concluded that a considerable fraction of the aerosol may be non-hygroscopic (gas/oil rags) in polluted air masses southeast of Tongliao. In fact, some aircraft and ground-based observations of the ratio of N _CCN to N _a over the Beijing region have indicated that there was indeed a large fraction (over 50%) of aerosol particles with small sizes (R _e<0.16 μm) (Lu and Guo, 2012); these particles could not be active at the 0.3% supersaturation rate, due to a large fraction of carbon in their composition, as well as their smaller size (Gunthe et al., 2011; Zhang et al., 2011). This may account for all the smaller activation efficiencies under polluted conditions. Additionally, the precipitation scavenging effect of soluble aerosols during long-range transportation can be one of the important factors influencing the activation efficiencies of aerosols (Kim et al., 2007).

Figure 10.  Relationship between CCN and aerosol (0.1 μm <d<3 μm) concentrations on 8 and 15 August (a) at 0.4% supersaturation and (b) at 0.6 % supersaturation.

During 2010-11, a set of vertical distributions of aerosol as well as CCN spectra were measured using airborne instruments in East Inner Mongolia, China, to investigate the potential influence of aerosols from heavily polluted megacities, such as Beijing and Tianjin, on the downwind area. The related conclusions are as follows:

(1) An analysis of the aerosol number concentrations indicated that the Tongliao region was much cleaner than heavily polluted regions, such as Beijing. The average vertical distributions of aerosols were consistent with that of the clean continental case. The mean surface level number concentration of aerosol particles (0.1 μm <d<3 μm) was approximately 1700 cm^-3, and the mean vertical profile approximately satisfied an exponential declining function with a scale height of approximately 1763 m.

(2) The monthly variation in the average aerosol number concentrations within the BL (height <1500 m) peaked in May (2455 cm^-3, 1868 cm^-3) and reached a minimum in September (684 cm^-3, 551 cm^-3) in both years. The monthly variation of CCN number concentrations at different supersaturations had the same trend as that of aerosols. The HYSPLIT model was run to determine the possible air mass influences on each measurement. The parameters c and k of the empirical function N _CCN=cS^k within the BL were 539 and 1.477 under clean conditions and 1615 and 1.42 under polluted conditions.

(3) The vertical distribution of CCN spectra, R _e, and aerosol number concentrations on 8 August 2010 and 15 August 2010 were analyzed separately. The R _e profile under polluted conditions revealed an insensitivity of the effective radius to altitude. Within the narrow range of 0.16-0.20 μm, the average value was 0.17 μm. In contrast, under clean conditions, the R _e profile varied at different levels, and the average value of R _e was larger than that under polluted conditions at all levels. The PSDs revealed different shapes of size distributions under both the clean and polluted conditions, especially for particles with diameters smaller than 0.5 μm. At all levels, the CCN spectra were consistent with an empirical curve of the form N _CCN=cS^k, and the empirical coefficient k in this study was much larger in comparison with that in most other areas of world. The coefficient k of most empirical functions was greater on 15 August 2010 than that on 8 August 2010 at the same levels.

(4) Measurements of aerosol particles, CCN spectra and back trajectories of aerosol simulated by the HYSPLIT model suggested that the atmosphere over the Tongliao region is often influenced by polluted air masses transported from Beijing and the Tianjin area. The aerosol (d>0.1 μm) concentration was as high as 2500-3000 cm^-3 in the BL under the inflow of pollutants. The ratio of N _CCN to N _a was 0.93 under relatively clean conditions, but decreased to 0.74 at 0.4% supersaturation in the anthropogenically influenced air.