Journal of Geophysical Research: Atmospheres

Importance of dust particles in the free troposphere over the Taklamakan Desert: Electron microscopic experiments of particles collected with a balloonborne particle impactor at Dunhuang, China



[1] Measurements of aerosol morphology and chemical elements were made in August 2002 at Dunhuang (40°00′N, 94°30′E), China, on the basis of direct sampling of free tropospheric aerosols with a balloonborne particle impactor, to understand nature of atmospheric particles over the desert areas in the Asian continent. Electron microscopic experiments of the particles directly showed that mineral (dust) particles were major constituents of coarse mode particles in the free troposphere over the Taklamakan desert. Typical types of the particles, according to energy dispersive X-ray (EDX) analysis, were Si-rich and Ca-rich particles in heights of about 3–5 km, and the ratio of those particle number to total particle number was about 0.71 in coarse mode range (diameter larger than 1.0 μm). The ammonium sulfate particles were major in fine mode range (diameter smaller than 1.0 μm). This result shows good correspondence with the lidar measurements, which were made in collaboration with this balloonborne measurements. The large depolarization ratio, according to lidar measurements, distributed from near the surface to about 6 km, suggesting that lots of particles having irregular shape (possibly dust particles) were in the free troposphere in summer over the Taklamakan desert. Trajectory analysis of air masses showed the possibility that westerly wind transported those dust particles (Kosa particles) to downwind areas even in summer season above about 5 km, which is interesting and useful information to give explanation on the aircraft measurements made at Japan, showing possible transport of dust particles in the middle and upper troposphere in summer season.

1. Introduction

[2] Recently many investigations on Asian continental particulate, especially Asian dust particles (Kosa particles; Kosa literally means yellow sand in Japanese), were made by a lidar [Kwon et al., 1997; Sakai et al., 2000; Murayama et al., 2001], a satellite [Imasu, 1992], a Sun photometer [Nakajima et al., 1989, 1996] and ground-based sampling of particulates [Quan et al., 1994; Okada and Kai, 1995; Zhang et al., 1998; Niimura et al., 1998; Zhang and Iwasaka, 2001], and numerical model [Denterner et al., 1996; Uno et al., 2001] suggesting possibility that Kosa particles contributes to geochemical cycle of atmospheric constituents and radiative balance in the east Asia and west Pacific regions.

[3] Long-range transport of Kosa particles, therefore, becomes a matter of great concern, and many campaigns such as ACE-Asia (Aerosol Characterization Experiment-Asia 2001–2002) including various type of field observations have been conducted. Many lidar measurements made in Japan suggested that long-range transport of Kosa particles was extremely active in the free troposphere over east Asia and west Pacific regions in spring [Iwasaka et al., 1988; Kwon et al., 1997; Sakai et al., 2000; Murayama et al., 2001], and the Taklamakan desert is suggested as possible one of strong sources of Kosa. Some investigators suggested that weak Kosa events which were so small that we could not detect near the ground, in addition to severe Kosa events, made possible contribution of geochemical cycle of minerals. Aircraftborne measurements also confirmed effect of weak Kosa events on chemical composition of particulate matter in the free troposphere over Japan [Mori et al., 1999].

[4] More recently weak Kosa events were suggested even in summer when few effects of Kosa on atmospheric aerosols over Japan and the Pacific ocean had been believed since Pacific high covers west Pacific region and largely weakens west wind in summer season by many investigations [Matsuki et al., 2002; Trochkine et al., 2002]. Electron microscopic experiments of the particles collected in the free troposphere over Japan suggested possibility that Kosa particles were transported on long-range from the Asian continent by westerly wind in the middle and upper troposphere (higher than about 4 km) even in summer. Therefore it is interesting, in order to understand long-range transport of dust particles in summer season, to clarify nature of aerosols over desert areas of the Asian continent in summer, especially in the free troposphere [Matsuki et al., 2002; Trochkine et al., 2002].

[5] However, concerning with nature of dust particles in the free troposphere over the desert areas of the Asian continent in summer, are little information have been suggested. Here we present results of electron microscopic experiments of particles collected with the balloonborne particle impactor in the free troposphere in August of 2002 at Dunhuang (40°00′N, 94°30′E), China, which is in east side of the Taklamakan desert and discuss importance of dust particle existence in the summer free troposphere.

2. Measurements and Results

[6] Aerosol particles were directly collected, to examine electron microscopic experiments of the collected particles, in the free troposphere with a balloonborne particle impactor in August 29, 2002 at Dunhuang, China (40°09′N, 94°41′E) (Figure 1). The Dunhuang city located in the east side of the Taklamakan desert which has been suggested as one of important sources of Kosa particles, and this location is considered to be effective to watch initial conditions of long-range transport of Kosa aerosols which is strongly affected by activities of west wind.

Figure 1.

Observation site, Dunhuang, China.

[7] Figure 2 shows the block diagram of aerosol sampler used here. Three low-pressure impactors with two stages were mounted on the balloon, and each impactors were operated in the three height ranges of 3–5 km, 5–7 km, and 7–8 km, respectively. The on-off of those impactors were automatically operated according to command signal from pressure sensor. Total weight of the sampling system including battery was about 9 kg, and volume 31 × 31 × 16 cm. The diameter of jet of first stage and second stage were 1.3 mm and 0.4 mm, respectively, and 50% aerodynamical cutoff size of particles were 1.36 μm (first stage) and 0.10 μm (second stage), respectively under flow rate of 1.8 × 103cm3/s and atmospheric pressure of 1013 hPa. The flow rate is expected to be the same in the free troposphere according to the laboratory test using vacuum chamber [Tuchiya et al., 1996]. Aerosol particles were directly collected on surface of electron microscopic nickel grid covered with nitrocellulose (collodion) film were precoated by carbon, and the grid was set on surface of stage of the impactor.

Figure 2.

(top) Block diagram of aerosol impact sampler system. (bottom) Schematic picture of the low-pressure impactor and airflow (arrows). The d1 and d2 is diameter of jet nozzle of first and second stage, respectively.

[8] The balloon train is shown in Figure 3, and rolled wire was extended to about 20 m during aerosol particle sampling. The Global Positioning System (GPS) was used to monitor the location of balloon and to search out the landing position of the impactor which was cut off from rubber balloon after finishing the particle collection.

Figure 3.

Balloon train used for aerosol collection; 29 August 2002, Dunhuang, China.

[9] Balloon sounding was made 29 August 2002 0323 (GMT) and weather conditions of that day are summarized in Table 1. Ascending speed of the balloon was about 340 m/min. First aerosol sampling started on 0331 (GMT) at 3179 m, and after then second sampling and third sampling were made at 5172 and 6872 m, respectively. Time schedule of particle sampling is shown in Table 1. Flight course of the balloon is shown in Figure 4. Distance between balloon sounding site and landing position (desert area) of the impactor was about 11.7 km and it took about two hours to recover the impactor after landing. The quick recovering of the impactor makes it possible to avoid effect of contamination from surrounding air. Particulate materials recovered was quickly stored in plastic box in which air was fully dried. Electron microscopic experiment of those particulate materials was performed in Japan (Nagoya University).

Figure 4.

Flight course of the balloon deduced by Global Positioning System (GPS) data.

Table 1. Sampling Period and Meteorological Conditions
Time, GMTAltitude Above Mean Sea Level, mSampling Time, minWeatherWind

[10] Examination of single particle with an electron microscope equipped with EDX has been used as one of most appropriate methods [Okada et al., 1987; Iwasaka et al., 1988; Spurny, 1999], and here scanning electron microscope (Hitachi Co. Ltd., S-3000N) equipped with EDX (Horiba Co. Ltd., EMAX-500) were used. The EDX is able to quantitatively detect the relative weight and atom ratios of elements (Z ≥ 5 except nitrogen) in a single particle.

3. Elemental Composition of Collected Particles

[11] In Table 2, combination of chemical elements of the particles collected during the balloonborne measurements. Type of particles in Table 2 is given on the basis of the quantitative results of EDX analysis by following relation according to Okada and Kai [1995]: weight ratio of element X, P(X) = X/(Na + Mg + Al + Si + S + Cl + K + Ca + Ti + Mn + Fe).

Table 2. Combination of Chemical Elements of the Collected Particles During the Balloonborne Measurements
Coarse Particles3–5 km5–7 km7–8 kmFine Particles3–5 km5–7 km7–8 km
Na + S200Si + Mg + Al + Fe100
Na + Mg + S + Ca100Si + Al + K010
Na + S + Ca100Si + Mg + Al010
Na + S + As100Ca-rich141
Si-dominant401Ca + S030
Si + Al300Ca + Si101
Si + Al + Fe200S-rich493132
Si + Al + Fe + Mg300S-dominant473031
Si + Mg + Al010S + Si100
Si + Ca + Fe100S + Na + K010
Si + Al + K100S + Zn001
Si + Na + Al100S + K100
Si + Na + Mg + Al + Fe100K-rich300
Ca-rich1613K + S300
Ca-dominant700total number of particles examined543735
Ca + S302
Ca + Mg100    
Ca + Si311    
Ca + Mg + Si100    
Ca + Zr100    
S + Na100    
Zn + Cl200    
Mg + Ca100    
Total number of particles examined4525    

[12] Major component of coarse particles is mineral particles, and Si-rich particle and Ca-rich particle are defined using threshold values of P(Si) > 85% and P(Ca) > 85%, respectively. The particles containing Na were found sometimes.

[13] In the present observations Si-rich and Ca-rich particles were, respectively, about 36% and 36% of total coarse size particles collected in 3–5 km heights (Table 2). Number of coarse particles collected in 5–7 and 7–8 km heights region also showed that Si-rich and Ca-rich particles were major. However, total particle number collected in those region were extremely low and it is a little difficult to suggest, only from the present observation, Si-rich and Ca-rich particles to be major in coarse size range in heights of 5–7and 7–8 km.

[14] Number of Si-dominant, Si + Al, Si + Al + Fe, and Si + Al + Fe + Mg particles were noticeable in Si-rich coarse particles collected in 3–5 km and showed ratio of 25%, 13%, 13%, and 18%, respectively. Number of Ca-dominant particles was extremely large, about 44%, in Ca-rich particles of 3–5 km heights, and Ca + S particle and Ca + Si particle showed 19% and 19%, respectively.

[15] Okada and Kai [1995] showed, on the basis of EDX analysis of single particles collected in the ground atmosphere at Zhangye (39°52′N, 100°23′E), China, that mineral particles dominated in size range of 0.1–6 μm in spring and summer and most of them were composed of aluminosilicate. Bulk analysis of X-ray diffraction patterns, SEM-EDX spectrum and ion chromatograph of the particulate matter collected in the ground atmosphere in spring at Aksu (40°37′N, 80°44′E), Qira (37°01′N, 80°44′E), and Shapatou (37°27′N, 105°01′E), China, with an Andersen sampler showed that the dominant silicate minerals are quartz, feldspar, chlorite and biotite, and that dust particles contained evaporates such as calcite (CaCO3), halite (NaCl), thenardite (Na2SO4) and gypsum (CaSO4) [Yabuki et al., 2002].

[16] In Figure 5a., typical examples of SEM images and EDX spectra of coarse particles collected in heights of 3–5 km are shown. Particle (a) is the example showing only Ca peak (Ca dominant) EDX spectrum, and very similar spectrum was obtained by Okada and Kai [1995] and Zhou et al. [1996] who suggested that this kind of spectrum indicated main composition of particle to be calcite (CaCO3). The EDX spectrum of particle (b) is the example showing Na peak, and halite (NaCl) and thenardite (Na2SO4) are suggested as main candidate composition. Typical NaCl particles frequently have cubic shape and show Na and Cl peak in EDX spectrum. No strong peak of Cl was found in the spectrum of particle (b) and peak of sulfur was identified. It is, therefore, possibly suggested that the Na-dominant particles (b) is mainly composed of thenardite (Na2SO4). Particle (c) showed irregular shape and EDX spectrum showed noticeable peak of Al and Si (Si + Al particle). Similar spectrum, sometimes along with peaks of Mg, K, Ca, and Fe, was frequently identified from the particles containing aluminosilicate collected in the ground atmosphere over desert areas in China [Okada and Kai, 1995], in the ground atmosphere at Beijing during severe Kosa event [Zhou et al., 1996; Zhang and Iwasaka, 1999; Zhang et al., 2000], and in the free atmosphere over Japan during weak Kosa event [Trochkine et al., 2002]. It is reasonable to assume, considering those previous observations even though those were based on particle collection in the ground atmosphere without Trochkine et al. [2002], that particle (c) in Figure 5a is composed of aluminosilicate.

Figure 5a.

Electron micrograph of individual particles collected in the free troposphere between about 3 and 5 km over Dunhuang, China, 29 August 2002. Coarse particles a, b, and c are calcite (CaCO3), thenardite (Na2SO4) and aluminosilicate, respectively (see text).

[17] Most of fine particles, as shown in Table 2, contained sulfur, and ratio of S-rich particles to total fine particles was about 90% (3–5 km), 84% (5–7 km), and 91% (7–8 km). Additionally S-dominant particles were major of S-rich particles (more than 95% in all heights regions). Therefore it is suggested that S-dominant particle is major of fine particles in the free troposphere. Typical SEM image and EDX spectrum of those particles are shown in Figure 5b.

Figure 5b.

Both fine particles d and e are ammonium sulfate (see text).

[18] Sulfate particles are most possible candidate as S-dominant fine particles, since many observations suggested that sulfate particles formed secondary in the atmosphere are major component in the atmosphere (free tropospheric sulfate particles over Japan by Mori et al. [1999] and Kido et al. [2001]; background site measurements by Bodhaine et al. [1981] and Parrington and Zoller [1984]; marine atmosphere measurements by Parungo et al. [1986] and Savoie and Prospero [1989]). The SEM image of most of sulfur-dominant particles, as shown in Figure 5b, has few small dots surrounding particles (no satellite structure). It has been suggested that sulfuric acid particles have satellite structure in electron microscope image and ammonium sulfate particles have no satellite structure [e.g., Bigg et al., 1974; Ono et al., 1983]. Therefore it is suggested that the S-dominant particles shown in Figure 5b are possibly ammonium sulfate ((NH4)2SO4) particles and not sulfuric acid droplets.

[19] Most of the fine particles collected here are assumed to be ammonium sulfate particles combining from SEM image and EDX spectrum, and a few particles only show satellite structure, which can be recognized as typical electron microscopic image of sulfuric acid droplet, on carbon thin film. In Figure 6 detection frequency of chemical elements of all particles collected in the balloonborne measurements are summarized on the basis of SEM-EDX analysis without the coarse mode particles collected in 5–7 and 7–8 km regions.

Figure 6.

Detection frequencies of elements for particles collected in the free troposphere over Dunhuang, China.

4. Discussion and Summary

[20] The measurements strongly suggested that coarse dust particles diffused in not only the boundary atmosphere but also the free troposphere over the Taklamakan desert area. Okada and Kai [1995] suggested mineral particles even in the summertime atmosphere over the desert areas of China since mineral particles were dominant (98% by number) in particle size of 0.1–5 μm radius in the ground atmosphere near Zhangye, China, which is downwind areas of the Taklamakan desert. According to aircraft measurements over Japan Kosa particles were sometimes detected in the middle and upper free troposphere but not in the low free troposphere [Matsuki et al., 2002; Trochkine et al., 2002]. In summer atmosphere seems to be relatively stable comparing with springtime atmosphere in the Taklamakan desert. However, very small storms with the scale of a few hundred meters; a few kilometers are often observed, and such small-scale disturbance can diffuse dust particles to the atmosphere.

[21] Most of previous measurements suggested that long-range transport of Kosa particles was very rare in summer season since strong Pacific high-pressure system occupied east Asia. However, those measurements made on the basis of the ground-based measurements and few measurements were made in the free troposphere. Therefore detection of Kosa particles in the middle and upper troposphere is very interesting and aerosol situation over desert areas of the Asian continent, especially free tropospheric aerosol condition, is desired to clarify.

[22] Figure 7 showed depolarization ratio deduced from the lidar measurements which was performed corresponding to the balloonborne experiment at Dunhuang, China. The large depolarization values suggested that nonspherical particles, possibly dust particles having usually irregular shapes, diffused from near the boundary to about 6 km height. This observational fact shows good agreement with the evidence that dust particles are identified in the free troposphere by electron microscopic (SEM-EDX) experiments.

Figure 7.

Lidar return: scattering ratio, R, corresponding aerosol mixing ratio (black lines), and depolarization ratio (percent), and δ, indicating nonsphericity of particles (gray lines). Definition of scattering ratio and depolarization ratio is given according to Kwon et al. [1997].

[23] Vertical profiles of particle number concentrations observed correspondingly by balloonborne optical particle counter 27 August 2002 (it was 2 days before the particle collection with the balloonborne impactor) (Figure 8). It is useful, even two days difference in observation time, in order to understand situation of particle distribution over the Taklamakan desert, to compare number concentration with particle type deduced from the electron microscopic experiment. The detailed description of the optical particle counter was given by Hayashi et al. [1998] and measurements, on the basis of balloonborne optical particle counters at Dunhuang, China, are shown by Kim et al. [2003]. Concentration of the particles with diameter larger than 1.2 μm was about 1 particle/cm3 in the free troposphere and roughly constant from about 1 km to about 6 km suggesting that dust particles was well mixed to about 6 km. There, concerning with free tropospheric aerosol concentration, are very few data which can be compared with the present observations. According to Zhou et al. [1994], concentration of particles with diameter larger than 0.8 μm was about 0.2–0.5 particles/cm3 in the free troposphere in summer over Beijing area. They measured number concentration sizing at only two diameters, 0.4 and 0.8 μm and it is impossible directly compare their values to the results of Dunhuang, but it is suggested that concentration of particles with super micron size (certainly dust particles) over the Taklamakan desert is apparently larger than the values at Beijing.

Figure 8.

The particle number concentration measured with an optical particle counter at Dunhuang, China, 27 August 2002.

[24] Many investigation have suggested the Taklamakan desert and Gobi desert as important sources of Kosa; lidar measurements [Iwasaka et al., 1983; Murayama et al., 2001], model simulations [Uno et al., 2001], and meteorological data analysis [Merrill et al., 1989; Xiao et al., 1997; Carmichael et al., 1998]. More recently it is suggested, from aircraft borne measurements over Japan, that air masses originated from the Asian continent contained lots of mineral particles in the middle-upper troposphere (not the lower troposphere) even in summer [Matsuki et al., 2002; Trochkine et al., 2002, 2003]. In summer the maritime atmosphere usually expanded over Japan as Pacific high pressure largely grow up, and it is believed that few Kosa particles are transported from the Asian continent. However, those observations suggested that there is possibility of long-range transport of mineral dust particles in the summer middle-upper troposphere from the Asian continent to eastern Pacific region.

[25] Trajectory of the balloon in Figure 4 showed that weak east wind dominated from the surface to about 4 km and wind direction largely changed to west (a little north) wind above 4 km. North and or east wind, according to analysis of long term observational data, are suggested to be dominant near surface in the Tarim basin [Sun et al., 2001; Sun, 2002]. They, from case studies of storm, suggested also important contribution of prevailing west wind on long range transport of dust particles above about 5 km over the Taklamakan desert.

[26] Backward and forward trajectory of air masses observed in 29 August 2002 were analyzed with the computer model, HYSPLIT, provided from NOAA (Figure 9), suggesting that some air masses are strongly affected by west wind and possibly transport mineral dust to downwind. The analytical results showed good correspondence with observational facts showed by Matsuki et al. [2002] and Trochkine et al. [2002, 2003] who suggested that air masses containing Kosa particles diffused to the free troposphere over Japan by west wind. The present observation showed only case studies, and therefore it is necessary to made more observations and understand fully representative aerosols in summer time and other seasons.

Figure 9.

(a) Backward and (b) forward trajectory: HYSPLIT provided by NOAA.

[27] The present observations showed that:

[28] 1. Most of coarse particles were composed of mineral dust particles in the free troposphere in summer over the Taklamakan desert.

[29] 2. Most of fine particles were composed of ammonium sulfate particles in the free troposphere in summer over the Taklamakan desert.

[30] 3. Existence of free tropospheric dust particles is confirmed by also lidar measurements which was made as collaboration work.

[31] It can be, from the present observations, suggested the Taklamakan desert as possible source of Kosa particles which diffuse to downwind areas such as Japan islands in the free troposphere even in summer seasons. The effect of Kosa particles traveling long-range in summer seasons on geochemical cycles of minerals, sulfur, and others in the region of east Asia and west Pacific region becomes newly problem. The measurements described here is the first success of direct sampling of particulate matter over the Taklamakan desert but only case study. Further measurements, therefore, are desired.


[32] 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 Balloonborne Measurements of Atmospheric Ozone, Aerosols, and Others). Staff members of Dunhuang City Meteorological Bureau gave us kind and helpful technical support during balloonborne measurements.