Vertical changes of aerosol concentration and size in the free troposphere over the Asian desert areas were first observed using a balloon-borne optical particle counter (17 August and 17 October 2001 and 11 January, 30 April, and 27 August 2002) at Dunhuang (40°00′N, 94°30′E), China. The concentration of particles with diameter larger than 0.1 μm was more than 1 particle/cm3 in the free troposphere in all seasons. The particle number-size distribution in this free troposphere shows the possible effect of diffusion of soil particles from the lower atmosphere to the free troposphere, and a noticeable peak of particle concentration is found in the supermicron size range of the number-size distribution pattern not only in the boundary mixing layer but also in the free troposphere. Trajectory analysis of air masses corresponding to aerosol layers in the free troposphere shows that the background aerosols could be transported far away. The scattering ratio and depolarization ratio measured with a lidar in May and August 2002 suggest the existence of nonspherical particles, possibly mineral particles, in the high free troposphere about 5–6 km over the Dunhuang area and showed good agreement, concerning distribution of particulate matter, with the balloon-borne measurements.
 The free troposphere over east Asia has been recognized as the region where long-range transport of Kosa (“yellow sand event” or “yellow sand particle” in Japanese; Asian dust) particles, especially in the spring season, is extremely active. Asian dust particles (Kosa particles) become a great concern from the viewpoint of global warming by radiative effect [Nakajima et al., 1989; Sokolik and Toon, 1996] and of the geochemical cycle of atmospheric constituents by long-range transport in the east Asia and west Pacific regions [Dentener et al., 1996; Martin et al., 1989; Gao et al., 1992].
 However, there were only a few observations on size and number concentration of aerosol particles in the free troposphere over the desert areas on the Asian continent [Zhang et al., 1998]. Additionally, most of the previous measurements treated the Asian dust storm effect only in spring, and very few observations have been made in other seasons. Recently, some investigators indicated the significance of background Asian dust aerosols [Iwasaka et al., 1988; Sakai et al., 2000; Trochkine et al., 2002; Matsuki et al., 2002]. A prevailing westerly wind is always observed in the free troposphere over the east Asia-west Pacific region, and it is reasonable to speculate that the effect of particles originating in the desert areas of the Asian continent, of course, most clear in spring, is expected in all seasons more or less over the east Asia-west Pacific region. Therefore it is important to know the aerosol distribution over the desert areas not only in spring but also in other seasons in order to obtain a better understanding of the regional and global environmental effect of Asian dust storm particles.
 Taklamakan desert, together with Gobi, has been recognized as an important source of Asian dust particles, which are transported by westerly winds and frequently observed in spring on the Japanese islands [Iwasaka et al., 1983]. Figure 1 shows the location of a balloon-borne observational site on the campus of the Meteorological Bureau of Dunhuang City, located at the eastern edge of Taklamakan desert.
 Here we measured number-size distribution, its vertical changes, and concentration of atmospheric particles with a balloon-borne optical particle counter in all seasons from summer 2001 to summer 2002 at Dunhuang (40°00′N, 94°30′E), China. Lidar measurements were made corresponding to the balloon-borne measurements in the spring and summer of 2002 near the balloon sounding site to compare vertical distribution of ladar returns with the balloon-borne measurements.
2. Balloon-Borne Measurements of Aerosol Size and Number Concentration
 Particle size and number concentration were measured with the balloon-borne optical particle counter (OPC) developed by the atmospheric research group of Nagoya University and SIGMATEC Co. Ltd. This balloon-borne OPC system was originally developed to observe polar stratospheric clouds and largely modified to realize effective portability in 1993, taking the following things into consideration: (1) Reduce of weight of instrument including battery, if possible to smaller than 6 kg. (2) Forward scattering light is used to measure particle size and concentration because strong scattering light intensity can make it possible to detect easily scattering light. (3) The system must be able to measure aerosols under atmospheric conditions at low pressure (about 5 hPa). (4) The system must be able to endure changes of pressure from normal pressure to low pressure.
Table 1 shows the main characteristics of the OPC system used here, and the optical system of the particle counter is schematically shown in Figure 2.
Table 1. Main Characteristics of the Optical Particle Counter in This Study
35 × 20 × 35 cm
3 kg (including battery)
Wavelength of laser
Scattering direction of light
Velocity of air in the pump
Time interval of sampling
400 MHz (connecting Vaisara radiosonde)
 The counter contains a semiconductor laser as its light source, a light-scattering chamber, a photodiode as a detector of the forward scattering light by the particles, an air pump introducing aerosols into the light-scattering chamber, electric components, and a battery used as the electric power source [Tsuchiya et al., 1996; Iwasaka et al., 1998; Hayashi et al., 1998]. The OPC system, connected with a Vaisara MRS radiosonde, sends the signals of scattered particles with the meteorological data to the receiver on the ground. The OPC used in these studies was designed to discriminate between particles with radii of 0.3, 0.5, 0.8, 1.2, and 3.6 μm (five channels). Tsuchiya et al.  showed many considerations and experimental results about capability of this OPC system.
 The main studies about balloon-borne OPCs were made by D. J. Hofmann and his colleagues, who measured the aerosols over Wyoming, United States (1993). More than 200 times, balloons with an OPC were launched to understand long-term trends of the stratospheric sulfate aerosol layer during the observational period, from 1971 to 1990 [Hofmann et al., 1975; Hofmann, 1993]. From vertical profiles gained in their study, they could not only explain the features of stratospheric sulfate aerosols but also discuss tropospheric aerosols.
 The measurements were carried out 5 times from the summer of 2001. Figure 3 shows the distribution of particle size and number concentration on 17 August and 17 October 2001 and 11 January, 30 April, and 27 August 2002, which are averaged by a running mean of 700 m interval. The observable height regions of balloon-borne measurements contained not only the troposphere but also the stratosphere and balloon burst heights of 30 km. Tropospheric profiles only are used in this study. Balloon launching times, dates, ascending speeds, and weather conditions are shown in Table 2. The weather conditions of 17 August 2001 and 30 April 2002 are not clear, and there was a little rain on the 2 days before in April, while fine and stable weather conditions were observed in August 2002.
Table 2. Balloon Launching Times and Dates, Ascending Speed of Balloon, and Weather Conditions of Observational Periodsa
Ascending Speed, m/min
Here, LT, local time.
17 August 2001
clear and sometimes thin clouds; weak west wind
17 October 2001
clear; no wind
11 January 2002
clear; no wind
30 April 2002
clear and sometimes thin clouds; weak west wind
27 August 2002
clear; no wind
 On 17 August 2001, very thin clouds were identified during the period of balloon-borne measurement. From aerosol concentration and humidity profiles we can show that these clouds, with thickness of 400 m, were identified at 7.6–8 km altitude from the extreme enhancement of particles with diameter ≥3.6 μm (most of the particles had grown to particles with D ≥ 3.6 μm). An aerosol layer with peak height of 6.7 km was identified below the clouds. In the aerosol layer, concentration of supermicron particles was noticeable. In the case of 30 April 2002 the aerosol distributions were complex because of weather conditions that were not clear; some dry regions and high-humidity regions are found below 5 km.
Figure 4 shows the concentration of supermicron-size particles. The existence of supermicron-size particles is certified at high troposphere over 5 km in all measurements, not only in the spring season, in which most of Asian dust storm occur, the so-called Kosa season, but also in other seasons.
3. Lidar Measurements
 The lidar system used here consists of an Nd/YAG laser with two wavelengths (1064 nm, fundamental wavelength; 532 nm, first harmonic wavelength) and a 30 cm Cassegrain telescope (Table 3). However, only 532 nm lights were used in these observations. Backscattered 532 nm lights are divided into parallel and perpendicular components in the receiving system in order to gain the depolarization ratio.
Table 3. Main Characteristics of the Lidar System at Dunhuang
 The scattering ratio of atmospheric aerosol R(z) is defined as follows:
where Bm(z) and Bp(z) are the backscattering coefficient of atmospheric molecules and aerosol particles at altitude z, respectively. The value of equation (2) can be regarded as the mixing ratio of particulate matter at altitude z.
 Total depolarization Dt(z) is defined by
where B(z) is the total backscattering coefficient including molecular components and aerosols components at altitude z and ‘//’ and ‘⟂’ are parallel and orthogonal components of backscattered light, respectively. It is assumed that the total depolarization ratio Dt is mostly the same as the depolarization ratio of particle, since for single scattering by spherical particles, the depolarized return signal is 0. If the depolarization ratio of aerosol Dt is high, nonspherical particles exist in that atmosphere. It is widely known that dust particles have strong nonsphericity and show a high depolarization ratio [Iwasaka et al., 1998; Sakai et al., 2000], like Asian dust aerosols.
 Lidar measurements were made in April and August 2002. The vertical profiles of backscattering ratio and depolarization ratio of 532 nm wavelength observed using the lidar in April and August 2002 are shown in Figure 5. Aerosol layers of small backscattering ratio and large depolarization ratio were observed. Backscattering ratios with 532 nm wavelength light were less than 5, while those over Japan in Kosa events were more than 10 [Kwon et al., 1997], but the depolarization ratio was very high. This fact suggests that the shapes of particles are nonspherical and these aerosol particles were composed of mineral particles.
4. Discussion and Summary
 Number-size distribution patterns of particles found in the boundary mixing layer and the free troposphere from these measurements without clouds effects are shown in Figure 6. A noticeable peak in the supermicron size range of the size distributions in the boundary mixing layer suggests that there are strong sources of supermicron particles, possibly mineral particles from the ground surface. Additionally, noticeable peaks of supermicron range were frequently observed in the distribution function of number concentration-size of free tropospheric particles. This fact suggests to us that mineral particles in the free troposphere are important not only in dust storm seasons but also in other seasons over the desert areas. The importance of a weak Kosa event, not only dust storms and background Kosa, has been suggested, and the present observations strongly support this suggestion. Here we defined that background Kosa is Asian dust particles that exist in all seasons.
Kwon et al.  found weak Kosa events in the high free troposphere in the summer and autumn seasons over Japan using lidar data. Kim et al.  and Matsuki et al.  suggest that features of seasonal change in the free tropospheric area are affected by Asian dust aerosols over east Asia through long-range transport by strong westerlies. Their airborne measurements over central Japan revealed steady transport of dust in the lower to middle free troposphere (2–6 km) during spring, including days with no evident dust outbreak. Such dust aerosols found as background were observed even in summer in the higher layers (>4 km) under the influence of remaining westerlies. Uno et al. , from their modeling results, suggest that Taklamakan-originating Kosa play a more important role in long-range transport than that from Gobi in two main Kosa-originating Asia continental deserts. These facts have important meaning in considering the aerosol effects for estimating global radiative force.
 Wind speed and wind direction were also calculated from GPS systems launched together with the OPC, and backward and forward trajectories of air masses were analyzed with the software provided by the National Oceanic and Atmospheric Administration (NOAA) (HYSPLIT transport and dispersion model) to see the history of those air masses. Figure 7 shows the wind speed and wind direction at Dunhuang on 11 January and 27 August 2002. The results of backward and forward trajectory analysis for 3 days are shown in Figures 8a and 8b. These figures suggest that there were strong westerlies in the free troposphere at high altitude over the Asian area and the possibility of long-range transport of air masses containing mineral particles throughout the observational periods.
 A size distribution function of lognormal type is widely used, and here we use the lognormal distribution function (zero-order lognormal size distribution (ZOLD) function) as the fitting function to the measurements [Deshler et al., 1993]. The function is
where D is the particle diameter, Dm is the mean diameter, and σ is the standard deviation. The OPC system used in these studies is composed of five channels, so the trial-and-error method was used for the fitting function. We estimated the first ZOLD function and compared with observed data to fix the other function. In Figure 9 the expected ZOLD functions are compared with the number-size distributions observed at Dunhuang, China. The parameters of ZOLD functions are summarized in Table 4. It is worthwhile to point out that, concerning the functions of the second mode, the mean diameter of free tropospheric aerosols is in the range of ∼0.7–1.0 μm and shows similar values to those of boundary layer aerosols (∼0.8–1.0 μm). From the present examination of ZOLD function fitting, the importance of coarse-size particles in the free troposphere is suggested.
Table 4. Parameters of ZOLD Functions
Free Tropospheric Aerosol (6.5–8 km)
Mean Diameter Dm, μm
Standard Deviation σ
 Small-scale disturbances are frequently observed in the desert areas, but they are not reported by meteorological observatories as a disturbance or storm since they are extremely common. Those disturbances can be suggested as effective processes to transport dust vertically from the ground to the free troposphere. However, in the present measurements, particle size was divided into only five channels, and this channel number is not adequate since a very limited size range is covered. It is necessary to measure with more channels in order to decide the best fitting functions.
 As shown in Figure 5, particles were distributed from near the ground to about 6 km and rapidly decreased to about 1 at about 6 km. Interestingly, this altitude is the same level as that of mountains surrounding the Taklamakan desert aerosol layer. This fact suggests that the mechanism of Kosa events, their transport, and the presence of background Kosa have a relationship with the topography of this area and it is necessary to consider the effect of background Kosa in studying Asian dust.
 Balloon-borne lidar measurements made at Dunhuang showed possible contributions of soil particles in the free troposphere, and regional transport of air masses containing coarse-size particles was important. Vertical profiles of aerosol particles from observational data in the free troposphere over Dunhuang strongly suggest that mineral particles are present that seem to have originated from desert areas by local influence and to have been transported far away in all seasons. However, there were not enough measurements to clearly understand the nature of particles in the Asian desert atmosphere. Many investigators have suggested that there is a strong seasonality in the wind system transporting mineral particles from desert areas on the Asian continent to the Pacific Ocean and in the source strength of soil particles from the boundary to the atmosphere [Merrill et al., 1989; Gao et al., 1992; Chung and Yoon, 1996; Zhang and Iwasaka, 2001]. Therefore it is desirable to make systematic observations covering all seasons in source regions of soil particles on the Asian continent.
 This investigation was supported by Japan Ministry of Education, Culture, Sports, Science and Technology (Grant-in-Aid for Specially Promoted Research, 10144104), and Japan Society for the Promotion of Science (Inter-Research Centers Cooperative Program, Stratospheric Physics and Chemistry Based on Balloon-borne Measurements of Atmospheric Ozone, Aerosols, and Others). Staff members of Dunhuang City Meteorological Bureau gave us kind and helpful technical support during balloon-borne measurements.