On the Association between Fine Dust Concentrations from Sand Dunes and Environmental Factors in the Taklimakan Desert

Abstract

Dust in sand dunes is an effective and important source of dust emission. The Taklimakan Desert (TD) is one of the main sources of global dust: the sand dunes account for approximately 85% of the total area of the TD. However, the dust concentration, emission characteristics, and physical factors of different parts of the sand dunes in the TD during the day and night, as well as dust and non-dust days, remain unclear. In this study, dust observations were collected over a 3 month period in the TD to investigate the physical processes by which dust moves across a surface and generates PM₁₀ and PM_2.5 from the top and bottom of sand dunes. The results showed that the daily average maximum concentrations of particulate matter with diameters below 2.5 and 10 µm (PM_2.5 and PM₁₀) in the dune during the observation period reach ~90 and ~190 µg·m⁻³, respectively. Dust emission generated in the saltation process (maximum emission of PM₁₀ was 3–5 mg·m⁻²·s⁻¹) in the TD dunes was larger than that in other areas and had a good correlation with u_* (friction velocity), where u_* = 0.4 m·s⁻¹ was the threshold of sand dune dust emission. The wind speed at the top of dunes was larger than that at the bottom, which was conducive to the accumulation of PM₁₀ at the top of the dune. Furthermore, the MLH (mixed layer height) decreased after sunset and the turbulence weakens, which was conducive to the retention of dust in the air. Moreover, the dust made the solar radiation at the top of the dune −15 W·m⁻² (average) lower than at the bottom. These results provided a new understanding of dune emissions in the TD and could be used for dust modeling in regional and global models.

1. Introduction

Dust aerosols affect the radiation systems between the earth and atmosphere, can alter albedo and rainfall patterns, and transport allergens and pathogens. Therefore, dust aerosols have a significant impact on climate change and human health [1,2,3,4,5]. Despite these important effects, current climate models have large uncertainties in dust emission rates and spatial simulations due to a poor understanding of the dust emission capacities and physical processes of different soil types [6,7,8]. Moreover, most parameterizations of dust emissions only consider emissions through soil aggregate fragmentation [9,10,11,12]. Although most dust models do not view sandy soils as an important dust source, it has been thought that reactive sand (loose sand grains) may produce dusts such as clay (PM_2.0) and silt (PM_2.0–PM₆₃) [13,14]. An additional PM₁₀ source in sandy soils is the abrasion of iron oxide coating on the particles [14]. Moreover, studies have shown that there is a high correlation between dust emissions and wind intensity [13,15]. The mechanisms of dust emission are mainly wind erosion and the saltation of sand particles [9].

Surprisingly, studies have shown that dunes may be a globally relevant source of dust. Sand dunes have higher erodibility than other soil types [13,15,16]. Approximately 20% of arid areas are covered with sand dunes globally, and approximately half of these dunes are considered active [17,18]. Moreover, most of a typical active dune is composed of sand particles ranging from 63 to 2000 μm, and the dust emitted by the dune is significantly finer than that from non-sandy soil [19,20]. Although satellite remote sensing can determine the source of dust uncertainty, satellites cannot obtain the information of small-scale dust concentration, land surface dynamics, and wind stress magnitude; hence, it is necessary to observe the dust of active dunes and the physical principle of dust emission.

Particle size of dust in sand samples (<63 μm) affects PM₁₀ emissions [21]. PM₁₀ concentration increased with the increase in initial sand content [21]. Aerodynamic dust entrainment is less efficient and can be significant if there is an adequate supply of loose dust particles at the surface [22]. Existing dust emission parameterized ranges are designed to represent simulations of the physics of the emission process, and sediment supply availability may be key to improving large-scale simulations of the dust cycle [23,24]. PM₁₀ emissions in the hinterland of TD can reach 1.6 kg day⁻¹ m⁻² in spring [25]. Owing to the thermal effect of the desert, a “deep” atmospheric boundary layer with 3–5 km height forms in the TD during the summer. It is also the floating dust residual layer of dust aerosol “stuck in the air” which can reach an altitude of 4 km [26,27,28,29]. Dust is “stuck in the air” for more than 200 days every year, and the occurrence period can span the entire spring and summer. During clear days, the dust concentrations are significantly lower than that on dusty days [29,30]. The dunes are active, complex, and varied in the TD. The wind-blown sand landforms are of different types: mainly linear, high, or compound longitudinal sand ridges. Many researchers suggest that dust sources are as diverse as dry lakes or dune fields [31,32,33].

Different types of dunes have developed in the desert, leading to the uneven distribution of surface energy, water, and the motion state of the atmospheric boundary layer, thereby increasing the uncertainty of near-surface parameters in numerical simulation and limiting the accuracy of weather forecasts and climate change predictions [34,35,36]. Topography can change synoptic-scale advection, thereby producing planetary waves, affecting mesoscale motion, and changing microscale turbulent motion [37,38,39]. Meteorological parameters, such as wind speed, temperature, and humidity, are the basic variables that characterize atmospheric processes. Related research results show that the temperature gradient can be approximated on flatter ground but changes on complex terrain [40]. Moreover, given that the relationship between the nighttime surface radiation temperature and topographic curvature yields a closer approximation than is estimated from the surface height, the slope has a significant effect. Therefore, the temperature on a steeper slope is also higher [41]. A weak topographic curvature change also affects the surface heat budget [42]. The ground surface, as an important component of the whole climate system, has complicated aspects involving different scales and spatial distribution. For instance, surface albedo, land use/land cover, topography, soil moisture, and urban areas determine the complexity and intensity of atmospheric processes. Surface features also affect these processes, influencing weather and climate [43,44,45]. However, they have little impact on the change in surface water, heat, and material exchange caused by the fluctuation of the microtopographic structure of the same underlying surface. Dust emission is closely related to wind intensity and saltation. While aspects may be unclear, many studies and models focus on and highlight the importance of dune systems on dust worldwide. Therefore, it is necessary to measure the dust concentration, emission, and environmental factors of sand dunes to understand the role of sand dunes and active sand, as well as the weather and climate. However, dust concentration, emission, and environmental factors of sand dunes in the TD remain unclear.

To improve our understanding of the association between dust emissions from active sand dunes and environmental factors, we have for the first time measured emissions of an active sand dune under natural conditions in the TD. In this study, we compared dust concentrations observed on the top and bottom of dunes, using the first observations in the TD about sand dunes to answer the following questions: (1) What is the difference between the fine dust emission measurements on the top and bottom of the dunes in the TD? (2) What physical processes drive the dust emission differences between the top and bottom of the dunes? These results provide a new understanding of dust measurements for the active dunes in desert areas.

2. Materials and Methods

2.1. Study Area

Dust weather refers to a weather phenomenon in which strong winds drive a large amount of dust and sand from the ground, increasing air turbidity and significantly reducing horizontal visibility. Dust weather is divided into three categories: sandstorm, blowing sand, and blowing dust. According to the Guide to Meteorological Instruments and Methods of Observation (2010) and GB/T 20480-2006 [46], a sandstorm is a dusty weather phenomenon in which a gust front or other strong wind blows loose sand and dirt from a dry surface, reducing horizon visibility to less than 1.0 km. Blowing sand is a dusty weather phenomenon in which the wind blows the surface dust, increasing the air’s turbidity, and reducing horizontal visibility from 1 to 10 km. Blowing dust is a dusty weather phenomenon in which sand or soil particles are uniformly suspended in the atmosphere (mostly from other places or remaining in the air after the end of local sand-blowing or sandstorm weather); this makes horizontal visibility less than 10 km. Dust weather events are frequent in the TD, with annual sandstorms lasting more than 30 days, blowing sand events occurring approximately every 70 days, and blowing dust events every 100 days. Dust weather events occur frequently in spring and summer, averaging a total of 112 days [47].

The surface wind direction in the Tarim Basin is northeasterly throughout the year, and dust aerosol is prominent. The wind speed and the aerosol optical depth (AOD) value are the highest in spring and summer [48]. Active aeolian movements are occurring throughout the TD. AOD (https://lpdaac.usgs.gov/products/mcd19a2v006/, accessed on 1 January 2020) is a subset of the data based on the 550 nm band of the MODIS-MCD19A2 dataset. Data processing includes projection conversion, clip**, and other normal preprocessing. The AOD level in TD is high, especially in July and August; the AOD value is typically above 2.0, occasionally reaching 4.0. Moreover, the distribution range of high AOD was wider in July and August than in June (Figure 1).

Vegetation is extremely scarce in the TD, as the surface is primarily composed of very fine sands (shifting sand with a light weight, fine grain size, and strong mobility) [49]. The relative height of NNE–SSW- or NE–SW-trending ridges is 40–50 m; ridge width is 1–3 km, and the length is 2–5 km. There are crescent-shaped dunes and sand-dune chains on the front edge of the sand ridges. Approximately 5.3 Ma, mobile dunes prevailed in the Tarim Basin [50].

TZ station (83°39′E, 38°58′N; 1099.3 m.a.s.l.) is the only comprehensive observation experimental station of atmospheric science in the world established deep in the hinterland (>200 km) of the TD (Figure 2).

2.2. Field Experiments

From 1 June 2019 to 31 August 2019, we installed three sets of observation instruments in Tianqi, China. However, the instruments at the bottom of the shady slope of the dune failed, leaving only the observation data at the top and bottom of the sunny slope of the dune (Figure 2). The east–north direction of the top and bottom of the dune is windward, and the opposite direction is leeward (Figure 2). Meteorological parameters at a 2 m height were observed, including PM_2.5 and PM₁₀ concentrations, downward short wave radiation (short for solar radiation (SR)), air temperature (T), relative humidity (RH), wind speed (WS), and wind direction (WD) at the top and bottom of sand dunes. The sampling data were instantaneous values, and 5 min interval outputs were combined into single instantaneous data values.

The observation ranges of PM_2.5 and PM₁₀ were 0–500 μg·m⁻³; the resolutions were 1 μg·m⁻³; and the observation accuracy was ±10%. The observation ranges of SR, T, RH, WS, and WD were 0–1800 W·m⁻², −50–85 °C, 0–100%, 0–60 m·s⁻¹, and 0–360°, respectively; their resolutions were 0.1 W·m⁻², 0.02 °C, 0.05%, 0.01 m·s⁻¹, and 0.1°, respectively; their observation accuracy was ±5%, ±0.2 °C, ±3%, ±(0.5 + 0.03 v) m·s⁻¹, and ±3°, respectively. Before the observation, we calibrated three sets of instruments on the flat natural sand surface; therefore, the observation error was almost 0. Then, we placed them on the top and bottom of the dune to observe. At TZ station, the meteorological parameters observed at 0.5-, 1-, 2-, 4-, and 10-meter height included T, RH, WS, and WD. The 3-meter height EC system was installed to measure the friction velocity (u_*) (Figure 2).

The geometric diameter of PM₁₀ is most relevant to the impact of dust on weather and climate [4]. In contrast, the PM_2.5 concentration has a more relevant impact on human health [3]. The height difference between top and bottom of the dunes is 44.5 m. Based on observation, we created a profile of the high NNE–SSW-trending ridges (Figure 2). During the observation period, the site was free of any vehicular activity, completely devoid of vegetation, and characterized by rolling sand dunes.

At the experiment site and TZ, the proportions of particle sizes <63 µm, 6–500 µm, and >500 µm on the sunny slope of the 0 cm depth dune are approximately 0.9%, 91.2%, and 7.9%, respectively; at the bottom of the 0 cm depth dune they are approximately 2.5%, 97%, and 0.5%, respectively. However, the proportions of particle sizes <63 µm in the sunny slope and bottom of the 5–40 cm depth dune (approximately 0.5–1.8%) are slightly larger than the proportion of 0 cm depth. Similar results were found on the flat surface at TZ, i.e., the surface dust content was relatively small. The soil water content on the dune surface is low (0.05 g).

We selected different weather conditions for analysis: sunny day (the standard for a sunny day was 0% total cloud cover and no weather phenomena), floating dust, blowing sand, and sandstorm. When analyzing the micrometeorological characteristics of sand dunes, 14 June, 20 July, 28 July, and 10 August were selected to represent sunny, blowing sand, floating dust, and sandstorm days, respectively. We define a “non-dust day” as a day in which there is no dust weather phenomenon (sandstorm, blowing sand, and blowing dust), while a “dust day” is defined as a day in which any dust weather phenomenon is present.

2.3. Methods

In this study, the friction velocity (u_*) was based on the EC system at TZ station, according to Equation (1):

$$u_{*}={({\overline{u^{\prime }{w}^{\prime }}}^2+{\overline{v^{\prime }{w}^{\prime }}}^2)}^{\frac{1}{4}}$$

(1)

where $u^{\prime }$, $v^{\prime }$, and $w^{\prime }$ denote the three-dimensional pulsating wind speed.

The mixed layer height (MLH) was introduced by Nozaki [51], according to Equations (2)–(6):

$$H_M=\frac{121}{6}(6-F)(T-T_d)+\frac{0.169F(u+0.257)}{12f\ln (z/z_{0m})}$$

(2)

where H_M is the MLH (m); F is six levels of atmospheric stability (level 1 is “Strong unstable”, level 2 is “Unstable”, level 3 is “Weak unstable”, level 4 is “Neutral”, level 5 is “Relatively stable”, and level 6 is “Stable”); T − T_d was the dew point deficit (°C); u is the 10 m height average wind velocity (m·s⁻¹); z is the observed height (m), which is 10 m in this study; z_0m is the aerodynamic roughness length (m), which is 3 × 10⁻³ m at TZ; and f is the Coriolis parameter.

F was determined by the Richardson number (R_i), using Equation (3):

$$R_i=\frac{g\sqrt{z_1{z}_2}}{\overline{T}}•\frac{\Delta T}{{(\Delta u)}^2}•\ln \left(\frac{z_2}{z_1}\right)$$

(3)

where g is the gravitational acceleration (9.8 m·s⁻²), T is the mean absolute temperature of z₁ and z₂ (K), $\Delta T$ and $\Delta u$ are the differences in temperature and wind speed between z₁ and z₂.

The dew point temperature (T_d) is found from Equation (4):

$$T_d=\frac{b\times {\log }_{10}\frac{e}{E_0}}{a-{\log }_{10}\frac{e}{E_0}}$$

(4)

where e is the vapor pressure (hPa), E₀ is the saturated vapor pressure at 0 °C (6.1078 hPa), a = 7.69, and b = 243.92.

The Coriolis parameter (f) is defined by Equation (5):

$$f=2\Omega \sin \phi$$

(5)

where $\Omega$ = 7.29 × 10⁻⁵ s⁻¹; $\phi$ is geographic latitude.

The value of z_0m was determined by Equation (6):

$$\ln (z_{0m})=\ln (z-d)-\frac{\kappa u}{u_{\ast }}-{\phi }_m$$

(6)

where d is the zero plane displacement (m), which is 0 m at TZ; κ is the von Kármán constant, which is 0.4 at TZ; and ${\phi }_m$ is a correction term.

In this study, the dust concentrations were directly measured by instruments (see Figure 1).

To study a partial approximation of how much dust is moving across the near surface, the dust concentration is multiplied by the wind speed to further obtain the PM₁₀-sized fraction that is generated in the saltation process. In this study, dust and non-dust days are used to distinguish between transport and local generation of PM₁₀.

3. Results

3.1. Overview of Dust and Meteorological Environment

Figure 3 shows the temporal changes in the PM_2.5 and PM₁₀ concentrations, solar radiation, and surface meteorological parameters at the top and bottom of dunes in the TZ area from June to August 2019. Figure 3 shows that PM concentrations in July and August are higher than that in June, which is consistent with the distribution of AOD in Figure 2. During the observation period, PM_2.5 and PM₁₀ maximum concentrations were approximately 90 µg·m⁻³ and 190 µg·m⁻³, respectively, while solar radiation varied from 110 to 580 W·m⁻², wind speed varied from 1 to 5.6 m·s⁻¹, air temperature varied from 17 to 35 °C, and relative humidity varied from 3 to 84%. When dust events occur, PM_2.5 and PM₁₀ concentrations increase rapidly, and solar radiation, air temperature, and relative humidity decrease rapidly from high initial values. This suggests that dust events occur under strong winds and daily temperature conditions (the weather gets warmer, and the air temperature increases) [47].

During the summer observation period, the average PM_2.5 concentrations at the top of the dunes were smaller (1.6 μg·m⁻³) than those at the bottom; however, the average PM₁₀ concentrations at the top of the dunes were larger (3.0 μg·m⁻³) than those at the bottom. This was because wind speed at the top of the dunes was faster (0.9 m·s⁻¹) than that at the bottom (

Table 1. Monthly average PM_2.5 concentration, PM₁₀ concentration, SR, WS, T, and RH at the top and bottom of the dunes from June to August 2019 at TZ in the TD.

	PM2.5 /μg·m−3		PM10 /μg·m−3		SR /W·m−2		T /°C		RH /%		WS /m·s−1
	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom
June	8.8	10.6	24.1	22.6	241.9	254.7	25.3	25.2	32.5	33.4	3.3	2.4
July	13	14.4	31.9	28.7	253.5	268.8	29.7	29.5	22.7	23.4	3.4	2.5
August	17.8	18.8	40.4	36.1	194	211.4	29.2	29.0	24.7	24.9	3.2	2.4
Average	13.1	14.7	32.3	29.3	229	244.2	28.1	27.9	26.6	27.2	3.3	2.4

). Further analysis shows that the PM_2.5 to PM₁₀ ratio in the dust at the bottom of the dune simply differs from that at the top. We can also see that the PM₁₀ concentration at the top of the dunes was higher than that at the bottom, making the solar radiation at the top of the dunes lower (15.2 W·m⁻²) than that at the bottom. The air temperature at the top of the dunes was only slightly higher (0.2 °C) than that at the bottom of the dunes (Table 1).

As shown in

Table 2. Average PM_2.5 concentration, PM₁₀ concentration, SR, WS, T, and RH at the top and bottom of the dunes on the non-dust and dust days at TZ in the TD.

	PM2.5 /μg·m−3		PM10 /μg·m−3		SR /W·m−2		T /°C		RH /%		WS /m·s−1
	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom	Top	Bottom
Non-dust day	6.3	7.5	16.9	15.7	418	448.3	28.7	28.1	23.9	25.5	3	2
The difference between top and bottom (non-dust day)	−1.2		1.2		−30.3		0.6		−1.6		1
Dust day	15.9	17.7	38.7	34.9	378.9	402.9	27.9	27.9	27.7	27.8	3.4	2.6
The difference between top and bottom (dust day)	−1.8		3.8		−24		0		−0.1		0.8

, the average values of dust concentrations, SR, WS, T, and RH at the top and bottom of the dunes for non-dust and dust weather were consistent with those in Table 1. However, the difference between dust concentrations at the top and bottom of the dunes during dusty weather was larger than that during non-dust weather. The differences in SR, WS, T, and RH between the top and bottom of the dunes on dust days were smaller than those during non-dust weather, indicating that the influence of dust aerosol on meteorological parameters was also significant (Table 2).

3.2. Average Daily Variation in PM Concentrations and Atmospheric Conditions

Figure 4 shows the mean diurnal variation in dust concentrations during the summer at the top and bottom of sand dunes in the TD. The mean diurnal variation in the PM_2.5 concentration at the top (bottom) of the dune ranged from 8.9 to 21.6 (10.3–23.7) µg·m⁻³, and the mean values of the PM₁₀ concentration ranged from 23.0 to 50.9 (21.1–45.6) µg·m⁻³. The diurnal variation in the dust concentrations on the dunes was significant. The first peak of the dust concentrations appears after sunrise from 08:00 to 09:00, and the dust concentrations gradually increase after 16:00. The maximum dust concentrations at the top were 21.6 and 50.9 µg·m⁻³; they occurred at 19:30. At the bottom, PM_2.5 and PM₁₀ concentrations had maximum values of 23.7 and 45.6 µg·m⁻³, respectively, at 19:50 (Figure 4a).

Dust concentrations change synchronously at the top and bottom of the dunes. At 15:00, the dust concentration differences between the top and bottom are small; then, the dust differences between the top and bottom gradually increase (Figure 4a). Air temperature hits its daily at 15:00, and the MLH is at its largest; this is conducive to the diffusion of dust and its lowest concentration. After 15:00, the MLH gradually decreases, which is conducive to the accumulation of dust concentration and the gradual increase in dust concentration. Both PM concentration peaks (8:00–9:00 and 19:00–20:00) are related to the stratification transition, which is favorable for PM accumulation. The first peak is the transition from the unstable boundary layer to the convective mixed layer, and the second peak is the transition from the convective mixed layer to the stable boundary layer (Figure 4b).

In comparison, PM₁₀ concentrations at the top are higher than that at the bottom, and the average difference is 3.0 µg·m⁻³. The PM_2.5 concentration exhibits an opposite change pattern, where the bottom is higher than the top. The average difference is 1.6 µg·m⁻³, indicating that the difference between the PM_2.5 concentration at the top and bottom of the dune is smaller (Figure 4a). Further analysis showed that PM_2.5/PM₁₀ at the top of the dune was lower than that at the bottom, with an average of 0.40 at the top and 0.5 at the bottom of dunes.

As shown in Figure 5, the diurnal variation characteristics of the dusts during non-dust weather are significantly different from those on dust weather. As shown in Figure 5a, on non-dust days, the mean diurnal variation in the PM_2.5 (PM₁₀) concentration at the top of the dune ranged from 1.7 to 21.6 µg·m⁻³ (1.7–49.3 µg·m⁻³), while, at the bottom of the dune, the mean diurnal variation in the PM_2.5 (PM₁₀) concentration ranged from 2.9 to 22.0 µg·m⁻³ (7.4–42.3 µg·m⁻³). On dust days, the mean diurnal variation in the PM_2.5 (PM₁₀) concentration at the top ranged from 12.1 to 23.3 µg·m⁻³ (30.0–52.7 µg·m⁻³), while, at the bottom of the dune, the mean diurnal variation in the PM_2.5 (PM₁₀) concentration ranged from 13.2 to 24.5 µg·m⁻³ (26.6–47.0 µg·m⁻³).

The difference between the daily peak values of the dust concentrations during non-dust and dust weather was small, and the difference was larger in the minimum dust concentrations. During non-dust weather, the PM_2.5 (PM₁₀) concentration was approximately 5 µg·m⁻³ (10 µg·m⁻³). After 17:00, with the decrease in the boundary layer height, the PM_2.5 and PM₁₀ concentrations increased rapidly; this high concentration lasted until midnight. The dust concentration was low around noon. The small peak from 8:00–9:00 is the same as Figure 4. At night, the stable boundary layer is conducive to the accumulation of PM, which reaches its maximum during the stratification transition period. However, during dust weather, the dust concentration was higher throughout the day; the average PM_2.5 concentration was 2.7–3.3 times that of non-dust weather; and the average PM₁₀ concentration was 2.7–3.1 times that of non-dust weather.

We show the diurnal variation in PM and meteorological parameters by the difference between the top and bottom (top value minus bottom value) of the dune (Figure 6). During the observation period, the average PM_2.5 (PM₁₀) concentration at the top of the dunes was lower (higher) than that at the bottom during non-dust and dust weather (Figure 6a,b), making the solar radiation at the top lower than that at the bottom (Figure 6c). However, the PM_2.5 (PM₁₀) concentration at the top was slightly higher (lower) than that at the bottom while the sun was up (Figure 4, Figure 5 and Figure 6a,b). The difference in the PM_2.5 (PM₁₀) concentration between the top and bottom of dunes for non-dust (dust) weather was greater than that for dust (non-dust) days (Figure 6a,b). The high concentration of PM₁₀ at the top of dunes is closely related to the wind speed at the top of dunes being higher than that at the bottom. As shown in Figure 6d, the wind speed difference between the top and bottom of the dunes is the same during non-dust and dust weather. The wind blows the dust, and the dust affects the temperature and humidity by influencing the short-wave solar radiation (Figure 6e,f). During the day, the air temperature at the top of dunes was significantly lower than that at the bottom for both non-dust and dust weather; however, at night, the air temperature at the top of the dunes was significantly higher than that at the bottom during non-dust weather, and the air temperature at the bottom was lower than that during dust weather. Humidity changes opposite to air temperature. In other words, at night during non-dust weather, the air temperature at the bottom was lower than that on dust days, and the humidity at the bottom was higher than that for dust weather. Therefore, the diurnal variation amplitude of air temperature and humidity for non-dust weather was more obvious.

The solar radiation at the top of the dunes was lower than that at the bottom. The maximum solar radiation at the top of dunes was 749 W·m⁻² and 812 W·m⁻² at the bottom, with a difference of 63 W·m⁻² (Figure 7a). The higher solar radiation makes the temperature at the bottom of the dunes 0.5 °C higher than that at the top during the day; however, the opposite is true at night (Figure 7b). This differs from previous knowledge regarding valleys. The wind speed at the top of the dune is consistently higher than that at the bottom, which can increase turbulence and air mixing, hence lowering air temperature during the day.

Air temperature at the bottom is 0.7 °C lower than that at the top at night (Figure 7b), indicating that there is an obvious inversion phenomenon at night. The diurnal variation in relative humidity was opposite to that of air temperature, and there was a reverse humidity phenomenon at night (Figure 7c). Owing to the presence of temperature inversion and reverse humidity, PM_2.5 is not easily diffused. Moreover, when the inversion intensity is weak and the wind speed is high at night, the PM_2.5 and PM₁₀ concentrations are high. Figure 7d shows that the wind speed at the top of dunes is higher than that at the bottom during the day, with a difference of 0.3–1.6 m·s⁻¹.

3.3. Daily Variation in PM Concentrations for Dust Events

This study established four experiment and control groups, with different weather conditions: 1 June 2019, sunny day; 19 August 2019, blowing dust day; 4 July 2019, blowing sand day; and 4 June 2019, sandstorm day.

On the sunny day, the PM_2.5 and PM₁₀ concentrations were low, and the diurnal variation was not significant (Figure 8a). On the blowing dust day, the PM_2.5 (PM₁₀) concentration at the top of dunes ranged between 23 and 57 µg·m⁻³ (50–133 µg·m⁻³), while that at the bottom it ranged between 16 and 59 µg·m⁻³ (40–117 µg·m⁻³) (Figure 8b). Compared with the sunny day, the PM_2.5 and PM₁₀ difference between the top and bottom of the sand dunes exhibited an increasing trend (Figure 8b). On the blowing sand day, the PM_2.5 (PM₁₀) concentration at the top of the dunes was ~63 µg·m⁻³ (~147 µg·m⁻³), while that at the bottom was ~148 µg·m⁻³ (~70 µg·m⁻³) (Figure 8c). At 13:19 on 4 July, the blowing sand event began, and the diurnal variation trend of PM_2.5 and PM₁₀ increased significantly. With the continuous development of the dust event, the PM₁₀ concentration at the top of the dunes reached its maximum value of 147 µg·m⁻³ at 18:20. On the sandstorm day, the PM_2.5 (PM₁₀) concentration at the top of the sand dunes was between 6 and 142 µg·m⁻³ (19–328 µg·m⁻³), while that at the bottom was between 10 and 131 µg·m⁻³ (20–245 µg·m⁻³) (Figure 8d). The sandstorm occurred from 17:05 to 17:32 on June 4, and the diurnal variation curve for PM_2.5 and PM₁₀ exhibited great fluctuations before the sandstorm, with PM_2.5 and PM₁₀ concentrations rising sharply before reaching their maximum values (Figure 8d). Compared with the sunny and other dust days, the PM_2.5 and PM₁₀ differences between the top and bottom of sand dunes were greatest on the floating dust day.

3.4. Dust Emission and Mixed Layer Height

The MLH is the key parameter of dust dispersion and concentration. We investigated the relationship between the MLH and dust concentrations in more detail on the dust days (Figure 9). On dust days, dust concentrations are negatively correlated with the MLH; this can be described by a power function. This is consistent with Zhou et al. [30]. For a lower mixed layer, the PM_2.5 and PM₁₀ concentrations are trapped in the MLH [52]. However, the effect is not significant on sunny days, indicating that the influence of the MLH on dust concentration during sunny days is complicated. Below an MLH of 1250 m, there is significant variability in the concentration, whereas above that concentration, no trends were observed (Figure 9). However, the relationship between the MLH and PM does not satisfy the power function when the MLH is low. This indicates that the relationship between the MLH and the PM concentration is more complex at lower MLH [52].

3.5. Dust Emission Potential

In the observed dunes, the dust emission capacity generated in the saltation process increased with high u_* values (Figure 10). PM_2.5 and PM₁₀ dust emissions at the top and bottom of sand dunes exhibited a low start-up u_* (approximately 0.4 m·s⁻¹) (Figure 10). However, the maximum PM_2.5 (PM₁₀) dust emissions at the top and bottom of the dunes in TD were 1.8 (5.0) mg·m⁻²·s⁻¹ and 2.3 (3.0) mg·m⁻²·s⁻¹, respectively, lower than that of shrub dunes and slightly higher than transverse dunes [14]. High dust concentrations in the dunes were consistent with the high dust emission (Table 1). Moreover, the PM_2.5 and PM₁₀ emission that is generated in the saltation process during non-dust weather is approximately 2× larger than that of dust weather, indicating that the impact of direct aerodynamic entrainment and self-abrasion dust during non-dust weather should not be underestimated. Yang et al. [53] suggested the effect of saltation on dust aerosols has been ignored for non-dust weather in the past. However, Macpherson et al. [54] suggested that dust emissions on desert surfaces may be derived from aerodynamic resuspension rather than saltation.

There is no doubt that dust concentrations in the TD are high when dust weather occurs, but most of these concentrations are due to weather systems rather than local emissions. Of course, a little dust is emitted locally on dust days and has a strong relationship with wind intensity. On dust days, when the value of u_* is high, the dust reaction is active, and the PM₁₀ (PM_2.5) concentrations remain at a high level. This is promoted by continuous and strong wind erosion and saltation. The PM₁₀ (PM_2.5) concentrations decrease after the peak value of u_*. Although the saltation intensity of sand particles is large and continuous on dust days, the high humidity makes the dust emission capacity significantly lower than that on non-dust days.

When u_* = 0.55 m·s⁻¹, the geometric mean values of the PM_2.5 emissions generated in the saltation process at the top and bottom of the dunes in the TD were approximately 0.86 (non-dust day at the top), 0.38 (dust day at the top), 0.42 (non-dust day at the bottom), and 0.25 (dust day at the bottom). In contrast, the geometric mean values of PM₁₀ emissions generated in the saltation process were approximately 2.07 (non-dust day at the top), 0.89 (dust day at the top), 0.79 (non-dust day at the bottom), and 0.44 (dust day at the bottom). The non-dust day is thought to represent the local emission of dust, which is conducive to the accumulation of PM_2.5 and PM₁₀ when u_* is large. However, the dust day represents the transportation of sand and dust, which is not conducive to the accumulation of PM_2.5 and PM₁₀ when u_* is large.

The PM₁₀ emission generated in the saltation process of the TD is slightly higher than that of the dunes from the Mojave Desert, Tengri Desert, and Mu Us Sandy Land [54,55]. This is because the particles are extremely fine, and the soil is dry in the TD. However, Cui et al. [55] suggested desert dunes are unlikely sources of high PM₁₀ emissions, while shrub dunes and rivers are sources of dust emissions because desert dunes usually lack fine particles. Nevertheless, Sweeney et al. [14] identified that dunes are an important source of dust. We suggested that although the dust (<63 µm) is only 0.5–2.5%, the TD area is large, and there is an ample dust source. Therefore, incorporating observation is a priority when using dust source schemes [15].

4. Discussion

Measurements of undisturbed fine dust concentrations from active dunes in the TD are reported in this study (Figure 11). Several key differences were found in dust movement between the active sand soil at the top and bottom of the dune: the size and difference in PM₁₀ and PM_2.5 concentrations (Figure 4, Figure 5 and Figure 6); dust movement concentration under different weather conditions (Figure 8); the MLH during dust movement (Figure 9); and the dependence of dust flux and shear velocity (Figure 10).

Topography and wind speed are the main factors causing the differences in dust transport in the TD; dust aerosol is transported from the east to the TD [56,57]. Although the surface disturbance activity during non-dust weather is weaker than that during dust weather, its contribution to dust aerosol should not be ignored [53]. In this study, the concentration of dust in dunes for dust weather is significantly higher than that for non-dust weather; however, the dust emission from saltation during non-dust weather is higher than that during dust weather on dunes (Figure 10). Moreover, on non-dust days, high u_* is conducive to PM accumulation, while high u_* is not conducive to PM accumulation on dust days due to local emission and transport (Figure 10). The sediment transport is mostly concentrated in the 10 cm above the surface; the sediment transport over 20 cm has little change [56]. In addition, different parts of dust dunes in the TD have different sand transportation rates [56]. The intensity of aeolian sand transport in the TD area is mainly concentrated during the day, particularly from 13:00 to 16:00 in the afternoon [58]. Coupled with the reduction in the MLH in the evening, the dust concentration reached its maximum at approximately 20:00 (LT) in the TD (Figure 4). On sunny days, the diurnal variation in the dust concentration is not obvious, and it is slightly higher during the night than during the day (Figure 5). However, the daily variation in dust concentration was evident on dust days. On floating dust day, continuous dust emissions, strong saltation, and suspension caused PM_2.5 and PM₁₀ concentrations to remain at relatively high levels favored by continuous and strong saltation and suspension. On sand blowing and sandstorm days, dust concentrations began at a low level accompanied by strong wind erosion in the afternoon, when the dust concentration increased significantly. After the peak, the wind erosion decreased as the wind speed decreased, and the concentration decreased significantly (Figure 8).

During the sand dust process, the meteorological parameters at the top and bottom of the dunes changed when PM_2.5 and PM₁₀ increased. The dust process reduced the downward short wave radiation, reduced the local air temperature, and broke the original vertical variations in the day and night temperature and humidity. Moreover, during the dusting process, the wind speed increased, and the MLH decreased. As shown in Figure 12a,b, when the PM_2.5 (PM₁₀) concentration was approximately 20–140 (10–160) µg·m⁻³, the relative frequency of dust concentration was higher for dust weather than for non-dust weather; when the PM_2.5 (PM₁₀) concentration was lower than 20 (10) µg·m⁻³ and higher than 140 (160) µg·m⁻³, this result was reversed. As shown in Figure 12c, the dust effect increases the solar radiation probability below 200 W·m⁻². Stronger winds are needed at the top of a dust day compared to the bottom of a dust day (Figure 12d).

The major global sources of dust are non-sandy soils with large amounts of fine clay and silt particles. Active sandy soils produce significantly less dust than non-sandy soils [21,59]. The re-release of dust particles previously settled between sand grains, the removal of the clay coating attached to the sand grains, and the wear of sand particles by wind are the three main mechanisms of dust generation by saltation for active sand grains. Feldspar sand is more effective than quartz sand for wear-induced dust generation [59]. It is hypothesized that active sand bodies can produce large amounts of dust emissions through sand wear mechanisms [13,14]. The deposition and accumulation of 2–63 μm sand grains on the leeward side of the dune resulted in finer soil (loess), which was also caused by sand wear [60]. However, several studies have shown that under typical natural sand transport wind conditions, aeolian wear contributes little to dust generation from active sand grains, and the removal of the clay coating is the main mechanism, with dust reemission contributing [21,61]. Sarre and Chancey [62] suggested a mechanism by which particle size is segregated by saltation bombardment and encourages the coarser material to preferentially rise to the dune surface. Yang et al. [49] showed that the <63 μm particles in the TD increased with the increase in height, indicating that these particles were mainly transported by suspension. The 63–200 μm particles decreased with the increase in height, suggesting that >63 μm particles were mainly transported by saltation. We observed a surface dust content of approximately 0.5–2.5% and even less after the dust weather passed through in the TD. In view of previous research on dust emission mechanisms, we hypothesized that for non-dust weather, direct aerodynamic entrainment and self-abrasion dust drives sand emission in the TD. Meanwhile, during dust weather, dust emission is driven by wind erosion and saltation of sand particles, and there is a negative correlation between sand emission and MLH [30]. Owing to the large area of dunes and the surrounding mountains on the north, south, and west sides of the TD, we speculate that the dunes in the TD are both a dust source and a sink. Dust gets “stuck in the air” after surface dust is released into the atmosphere when the stable atmospheric stratification occurs [26,27,28,29]. The difference between the sand emission at the top and bottom of the sand dune is closely related to the intensity of wind speed and sand particle size, where the wind speed at the top of the dune is larger, and the sand particles are finer than those at the bottom of the dune. The difference between the PM₁₀ emission at the bottom of the sand dune is significantly lower than that at the top of the sand dune, making the solar radiation, temperature, and relative humidity different between the top and bottom of the sand dune.

The ratios of feldspar/quartz and calcite/quartz in the sand grains of TD are higher than those of other deserts [63]. For the cycle of temperature, humidity, and wind speed during the day and night in the desert, the sand grains of the dune collide and fall to form fine particles. Eolian abrasion and finer particle sizes of dune sediments are a possible result of the dust emission process in the TD. Moreover, the sediments provided by the TD are partially blown away and become the source of eolian loess in China [63]. During the sandstorm event, the dust emission flux was 27.2 ± 4.1 µg m⁻² s⁻¹. However, owing to the influence of the terrain and remaining wind direction, the contribution of TD dust transport was smaller than that of the Gobi Desert. However, under the influence of the westerly jet, a small amount of TD dust rose to more than 5 km and spread for a longer distance [28]. The dust emission generated in the saltation process of TD dunes on non-dust days cannot be ignored. However, research shows that the contribution of dune dust to high PM₁₀ concentration is low [64]. The contribution of dust to global emissions and transport is not clear, requiring further study.

5. Conclusions

Dust sources at the top and bottom of sand dunes are critical in their impact on dust emission contributions, environment, climate, and simulation. In our study, the dust concentration and emission potential generated in the saltation process at the top and bottom of sand dunes in the TD were studied using dust observation instruments. The relationship between dust emissions and the environmental factors was discussed. Our main findings are as follows:

• The difference between PM_2.5 and PM₁₀ on non-dust days was small, but it was large on dust days. The ratio of PM_2.5 to PM₁₀ at the top and bottom of the dune differs, making the PM₁₀ (PM_2.5) concentration higher (lower) at the top of dunes than at the bottom;
• Dust concentrations are higher in the afternoon, reaching their daily peak at approximately 20:00 (LT). On dust days, the average daily concentration of dune dust is approximately three times that of non-dust weather. However, the contribution of dune emissions generated during the saltation process to dust aerosol on non-dust days cannot be underestimated and is more than twice that of dust days. The mechanisms of dune dust emission in the TD are likely to be wind erosion and saltation (direct aerodynamic entrainment and self-abrasion dust) during dust (non-dust) weather;
• High dust emission, generated in the saltation process, and concentration correspond to high wind speed, high friction velocity, low humidity, and low mixed layer thickness. The difference in dust concentration at the top and bottom of dune is closely related to wind intensity. When the WS at the top and bottom of dunes is 4–10.5 m·s⁻¹ and 2–8.5 m·s⁻¹, respectively, the relative frequency of dust weather is higher than that of non-dust weather. With the increase in u_*, dust emissions increased, particularly on non-dust days. An MLH < 1250 m has a negative correlation with dust emissions and can be expressed as a power index relationship on dust days.