The Horizontal and Vertical Characteristics of Aeolian Dust from Riverbed

Arid riverbed is an important source of Aeolian dust to influence the atmosphere of nearby village or downtown. In general, Aeolian dust pertains to wind activity in the study of geology, environment and meteorology on particle suspension. This field study set the horizontal/vertical sampling system up in order to investigate the Aeolian dust dimensional distribution properties from the riverbed of the Jhuoshuei River. In addition, the study utilized the unmanned aerial vehicle (UAV) imagery to discriminate the covering condition of riverbed. The results revealed that the percentage of riverbed covering conditions, including bare zone, wetlands, green covering and water covering was about 50.1, 15.7, 16.8 and 17.4%, respectively, on January 5, 2011. Two Aeolian dust cases from riverbed were measured on November 26, 2010 and January 15, 2011. First sampling case was under slower wind speed, and the total particulate matter (PM) concentration in vertical sampling was almost decreased with increasing sampling height, however, the phenomena of second case (faster wind speed) was just on the contrary. Besides, Aeolian dust was distributed in a bimodal or a multimodal curve, and the main peak size was above 10 μm. Mode size of suspension particle diameter was increased as the wind speed increased (14.8 μm/ first case; 21.3 μm/ second case). The total mass concentration ratio of south site to source site (one meter height) was about 1:9 in first case and about 1:3 in second case. Comparing with those two cases, the faster the wind is, the shorter the surface roughness height is (3.08 mm/ first case; 1.07 mm/ second case). Besides, as the wind speed increasing, the friction velocity was also increased (0.33 m s/ first case; 0.68 m s/ second case). Consequently, this study clued the spatial variability of river dust events, which can further aid the site investigating, forecasting and preventing of dust influences.


INTRODUCTION
Aeolian dust is a kind of atmospheric mineral particle, including loess, sand and clay, which is suspended by Aeolian processes.These processes of particulate suspension such as raising, transportation, and deposition are important activities of the geomorphic surface (Offer and Goossens, 1995;Delmas et al., 1996;Field et al., 2009;Field et al., 2012).Especially, the finer and dryer particle are easily suspended by wind force in the arid or semi-arid areas (Goossens and Offer, 1997;Hahnenberger and Nicoll, 2014).
Aeolian dust emissions from semi-arid and arid surfaces ensue from two steps of Aeolian process.First, before dust suspension, particulate will be rolling (creep) and bouncing (saltation) when the friction velocity exceeds the threshold wind friction velocity (Bagnold, 1941;Saffman 1965;Greeley et al. 1977;Sharratt and Vaddella, 2014).Particularly, dust collides with bigger sand aggregates by inter-particle forces, in the meanwhile, the abrasion and splintering actions between sand and rock bring finer particles (Kok, 2011).Second, due to Bernoulli's principle, dust was directly carried by aerodynamic force and lift force (Kok et al., 2012), causing not only horizontal but also relatively vertical dust flux (mg m -2 s -1 ) for a higher given wind speed (Goossens and Offer, 2000).Sand suspension is not only affected by wind, but also influenced by water indirectly, including river flow and raining.It'll change the moisture content and surface structure of sediment (Karanasiou et al., 2014).Moisture content transforms the soil properties, like viscosity, size, unit weight and elements.And, soil crusting is a kind of surface structure change, which is caused by raindrop force impacting onto the bare zone.The raindrop falling down may seal the porous of surface soil forming sedimentation crusts.The friction velocity of crusting surface is faster than loose surface of the bare zone (WMO, 2008;Webb and Strong, 2011).
Significantly, the Aeolian dust emission is not only affected by the weather condition, but also related to bare zone size, covering condition, and influence time.Some researchers used different tools to know the bare zone size and covering condition, even in air quality measuring, such as unmanned aerial vehicles (UAV) and satellite (Bian et al., 2011;Zhao, 2012;Hahnenberger and Nicoll, 2012;Hugenholtz et al., 2012;Shen et al., 2015).The accuracy of display resolution from UAV was great then satellite imagery.UAV is a significant tool to observe the characteristic on the earth's surface due to the high resolution.
In recent years, the objective of research on dust suspension and transportation has grown rapidly.There are three kinds of method to measure the dust emission, including the theoretical derivation, numerical analysis (wind tunnel simulation), and experimental test (sampling and remote sensing) (Furieri et al., 2013;Modaihsh and Mahjoub, 2013;Margielewski et al., 2015;Vijayakumar et al., 2016).Some researchers measured the PM chemical composition, size distribution, or mass/number concentration in ambient air as well as in the source to identify the different PM periods (Liang et al., 2015;Upadhyay et al., 2015).The results indicated that the suspension particle occupies a vital character in the optical, physical, chemical, and biological effects in the atmosphere, such as visibility, solar radiation, and photosynthesis (McGowan et al., 1996;Erell and Tsoar, 1999;Margielewski et al., 2015).In general, pool visibility is essentially assigned to the extinction of visible light by particles (Friedlander, 2000).Several studies reported that the atmospheric visibility is related to PM (Pillai et al., 2002;Pui et al., 2014;Cheng et al., 2015).The higher the particle mass concentration is, the poorer the visibility is (Kim et al., 2001;De et al., 2005;Cheng et al., 2015).At the same time, dust emission also influences the regional air quality and bring about human health problems (Ono, 2006;Goudie, 2009;Rashki et al., 2013;Goudie, 2014), especially, the contaminants along with dust affect human health as its transport to the densely populated town or city.Not only the visibility, ambient air quality, and health problems, but also global climate effects are influenced by Aeolian dust (Bauer and Ganopolski, 2010;Evan et al., 2014;Lancaster, 2015).
Aeolian dust is frequently happening around our living environment, such as the road dust due to the vehicle moving, human activity or the wind blowing (Karanasiou et al., 2014).Besides, some dust comes from gravel processing site, construction site or dry fallow due to wind blowing (Chang, 2006).Moreover, the sandstorm, which comes from Desert or dry Lake, is most comment environment issues in East Asia, Middle East, North America, Australia and North Africa (Chen et al., 2002;Ono, 2006;Reynolds et al., 2006;Rashki et al., 2013;Zhu and Yu, 2014).It is the transnational materials transfer.The composition of the material particle from the long-term transport sand storm was mostly finer than 10 µm (In and Park, 2003;Pye and Tsoar, 2009).Furthermore, some local places were affected by the river dust that comes from the arid riverbed, which affect the nearby towns (Lin and Yeh, 2007;Vickery and Eckardt, 2013).
Thus far, the Aeolian dust from the Jhuoshuei River is always an important emission source in Central Taiwan (Chen et al., 2015).However, the dust emission features from the riverbed was scarcely investigated, especially for suspension particle size distribution and vertical concentration condition.In order to probe the spatial and temporal variability of Aeolian dust.This thesis tried to explore the characteristic of Aeolian dust emission from a semi-arid riverbed of Jhuoshuei River.

METHODS
The purpose of this measurement tried to provide insights into the relationships between source wind condition and Aeolian dust emission.Therefore, in this section, the study zone and the weather condition were introduced first.After that the UAV imagery analysis process was used to investigate the riverbed current situation.Besides, wherever possible, the wind speed, wind direction, particle size distribution and mass concentration of Aeolian dust were measured for all field sampling sites.

Study Zone and Weather
The exploration was conducted on 26 th November 2010 (first case) and 15 th January 2011 (second case) in Jhuoshuei River, West-Central Taiwan (Fig. 1).It is the longest river in Taiwan, which is covering 186 km in length.The gradient of Jhuoshuei River is 1/46 in average (Li et al., 2013), however, the gradient of the upstream tributary is on a very steep slope (Yang et al., 2011).Therefore, the erosion process of water is extremely strength due to the orogeny in Taiwan (Siame et al., 2010).The stratum of mountain surface is the sediment from the ocean, and it's young and also active in Taiwan.Therefore, sedimentary rock (shale and slate) is the mainly component of the upstream mountain here.If it's affected by erosion, such as rainfall, river flow or anthropogenic activity, sand and mud will flux along with rapid streams and then deposit in the flat area of the river (Li et al., 2013).
Wind force and moisture are two of the most important weather properties in sand suspension (King et al., 2011).Taiwan is located on the tropic of cancer, and its general climate is the tropical marine and also non-arid here, where the day length and temperature are fluctuating.The annual average rainfall is about 2,129 mm (from 1981 to 2015); the rainy season usually happens in summer as the Asian monsoon is coming.In addition, the typhoon is greatest severe weather from June to September (Fang et al., 2009).During the winter and early spring (November-March), central and southern parts of Taiwan are generally sunny and dry.Moreover, the wind energy in Taiwan is copiously due to the Asian northeast monsoon in winter and tropical cyclones (typhoon) in summer.Northeast monsoon is the most important effects of Aeolian dust in Taiwan.Besides, due to the strong sea breezes during the daytime (Pokhrel and Lee, 2011) in winter, the wind speed of monsoon is faster than in summer.
For that reason, this study focused on investigating the effect of Aeolian dust from downstream sandbank of Jhuoshuei River.Thus, the local weather condition (including wind speed and direction) was based on Taiwan EPA air quality monitoring station.As shown in Fig. 1, the background and downstream air quality monitoring station was Er-Lin station in Chang-Hua County and Lun-Bei station in Yun-Lin County, respectively, due to the northeast monsoon.Besides, the sampling not only located on the riverbed (source site) but also measured upon the north embankment (north site) and on the top of a community centre of the nearest south village (south site).

Covering Analysis of Riverbed by UAV Imagery and Computational Chromatic Analysis
In this part, the imagery of the Jhuoshuei River from UAV was analysed in this study that seems to be well suited for providing the covering information.The digital camera (EOS 500D with 50 mm prime lens, Canon) was installed in UAV which used to take the riverbed photograph at 1,200 m height.After that, the computational chromatic analysis was used in river covering and bare zone measurement.However, the original image from UAV was complicated.Thus, before chromatic analysis, the non-river course area was removed by Photoshop CS4.Afterward, the coverings and bare zone were dissected in computational chromatic analysis software "ImageJ".It can calculate area and pixel value statistics of user-defined selections.In brief, the percentage of bare zone, green covering, and water covering areas were calculated by the computational chromatic analysis.

Sampling
Aeolian dust emission is not only horizontal but also vertical transport that spread in three dimensions.Hence, in this thesis, the particulate matter (PM) measurement included horizontal and vertical sampling.The target of horizontal sampling was focused on the relations of the mass concentration increment between north site (upstream), source site (sandbank) and south site (downstream) at sampling site (as shown in Fig. 1).Besides, the vertical sampling, which was five meters height sampling tower, was set-up to collect the Aeolian dust sample from the riverbed at the source site.The PM mass concentration and size distribution were measured by using Marple personal cascade impactor (Model 298, Thermo Andersen Inc., USA) sampler.The aerodynamic diameter range of Marple sampler was from 0.5 µm to 21 µm with a final filter.In principle, the Marple sampler equipped the 34 mm PVC fibres filter.Those filters were conditioned in a conditioning room for 24 hours where the relative humidity (RH) and temperature was kept at 40 ± 5% and 20 ± 1°C, respectively.And then, all sampling filters were weighed by an ultra-microbalance (Model XP2U, Mettler-Toledo International Inc., Switzerland).
Moreover, due to the reason in above, this study tried to examine the influence of northeast monsoon on Aeolian dust from the riverbed.Thus, the ambient wind speed and direction detector (Weather Wizard III, Davis Instruments, USA) was not only installing on horizontal sampling site but also used to measure the vertical wind situation at 1 meter, 3 and 5 meters heights in the source site sampling.

RESULTS AND DISCUSSION
Aeolian dust from riverbed was an important source to affect the ambient air quality and the human health.The dust suspension from the riverbed was influenced by some properties, such as the bare zone size, moisture content, weather conditions and sediment properties.Therefore, the following section investigated the covering condition of the riverbed, and ambient air quality by the horizontal/ vertical sampling.

Covering and Weather Conditions of Jhuoshuei River
Covering condition of the riverbed was one of the most important properties on dust emission.Fig. 2 showed the downstream covering condition of the Jhuoshuei River on January 5 th , 2011 by using the UAV photography system.According to the computational chromatic analysis, the results indicated that the bare zone was about 50.1% of the total riverbed zone.Besides, the effective bare area was smaller due to the land surface crusting or something covering.The partial land of bare area was obviously covered with dry straws by farmers due to the preparation of watermelon planting.Furthermore, as shown in Fig. 2, there were two types of surface crusting, first one was thin slice (< 1 cm), and another was thick cake (about 3 cm in average).Those surface crusts protected the riverbed to reduce the effect of Aeolian process.
Along with Fig. 2, the green covering was about 16.8%.Therefore, the other area could be regarded as the water covering and wetlands near the water, however, the water covering and wetlands were difficult to recognize.Finally, this study changed the contrast and the RGB intensity of the image to discriminate the water zone and wetlands.According to this method, the water covering was about 17.4% and the wetlands was about 15.7% of the total riverbed.
In addition, the weather in the study zone was less rain and almost at dry condition from October 2010 to April 2011.The average value of monthly accumulated precipitation was only around 18.14 mm at these seven months (data from Yun-Lin weather station, Central Weather Bureau, Taiwan).Besides, the precipitous topography procures the short residence time of the river stream in Taiwan.Therefore, in measuring period, the riverbed lied in the semi-dry condition.The Aeolian dust might influence the nearby village ambient air quality severely during the windy weather situation.

Horizontal and Vertical Characteristics of Aeolian Dust
According to the above reason, this study investigated two event cases, the first case happened on November 26 th , 2010, and the second case was on January 15 th , 2011.The first case was the Aeolian dust emission in slower wind speed condition.Then the dust emission of faster wind speed condition was probed in the other case.The following results were discussing the particulate matter mass distribution and wind profile according to the vertical and horizontal sampling.
(A) First Case (November 26 th , 2010) In principle, the weather condition was an important factor of Aeolian dust suspension and diffusion, and it's also a vital project of ambient air quality.The stronger northeast monsoon happened on November 26 th , 2010.The wind direction was almost blowing from the north to the south as shown in Fig. 3. Wind speed is one of the most important factors in dust suspension.The hourly average wind speed was about 2.7 m s -1 and the maximum hourly wind speed was about 4.5 m s -1 in Lun-Bei air quality monitoring station (from 10 a.m. to 6 p.m.).Nevertheless, the average wind speed in the source site (five meters height, from noon to 5 p.m.) was about 6.03 m s -1 , and the maximum wind speed was 8.5 m s -1 , as shown in Fig. 3.In fact, the discrepancy of wind speed in the atmospheric boundary-layer was influenced by terrain, vegetation and  structure (Conan et al., 2015) due to the surface roughness length increasing (Tieleman, 2003).Besides, the accumulated precipitation from November 23 th to 26 th , 2010 was zero in order to obviate the influences of moisture.
In horizontal sampling, the PM size distribution and mass concentration were measured by using model 298 Marple sampler.As demonstrated in Fig. 4, the PM mass concentration in the source site was larger than other sites.The total mass concentration ratio of south site to source site (1 meter height sampling) was about 1:9.Fig. 4 also indicated that the mode of Aeolian dust in the source site was about 14.8 µm (coarse particle).However, the south site was affected by fine particle.In principle, the obstacle and covering drop the wind speed, therefore, suspension particle retard and deposit easily (Youssef et al., 2012).Especially, the vegetation covered riverbed increase the surface roughness and roughness height (Counihan, 1975;Tieleman, 2003), it change the Aeolian dust diffusion.According to these reasons, at slower wind speed condition, the downstream village's ambient air quality might not be affected by coarse dust even if the sediment suspension from riverbed.However, as the result of exposure risk studies, the particle size of primary pulmonary deposition was below 0.1 µm (Sahu et al., 2013;Patterson et al., 2014).It should pay attention to the effect of Aeolian dust on health even in slower wind speed condition.
Table 1 exposed the vertical sampling data which included PM concentration, size distribution, and wind speed in each sampling height.First, the total PM concentration almost decreased with increasing sampling height.Besides, in this case, the coarse dust was measured at one meter height principally.However, the fine dust almost distributed at two meters height.The results proved that the primary contribution of Aeolian dust was on coarse particle, additionally, the wind blow suspension height of fine particle was greater than coarse size particle.It verified that the coarse particle deposited in the vicinage by gravitational settling and aerodynamic drag force.Moreover, Table 1 revealed wind speed in each sampling height also.According to the equation of wind profile (Eq.( 1)), if the Von Kármán constant is about 0.4 (atmospheric condition, Zhang et al., 2008), the calculation surface roughness height and friction velocity was about 3.08 mm and 0.33 m s -1 , respectively, in this case.The surface roughness height was almost the same with WMO's report in 2008.In mud flats or snow with no vegetation and obstacle situation, the surface roughness height was about 5 mm (WMO, 2008).
where the u z is the mean wind speed at height z (m) above the ground; κ is the Von Kármán constant; u * is the friction velocity (m s -1 ); z 0 is the surface roughness (m).
(B) Second Case (January 15 th , 2011) The weather condition of the second case was also influenced by the northeast monsoon on January 15 th , 2011.The cardinal direction of wind was also coming from the north.Besides, rainless condition was also occurred for three days long before sampling, in this case.According to Lun-Bei air quality monitoring station, the hourly average wind speed from 10 a.m. to 6 p.m. was about 8.3 m s -1 and the maximum hourly wind speed tended to 11.0 m s -1 .In addition, the average wind speed in the source site (five meters height, noon to 3 p.m.) was about 14.0 m s -1 , and the maximum wind speed was around 16.9 m s -1 , as presented in Fig. 5.The wind speed of this case was higher than the first case undoubtedly.
Fig. 6 exhibited that the particulate matter size distribution and mass concentration of horizontal sampling in the second case.PM mass concentration of south site was progressively decreasing in the wake of that the dust cloud was crossing a length of distance, however, it's also higher than the north site.The total mass concentration ratio of south site to source site (1 meter height sampling) was about 1:3.And the particle mass size characteristics of Aeolian dust were distributed in a bimodal or a multimodal curve (Dobrzhinsky et al., 2012).The available evidence giving the same information with other studies, coarse particle with the size greater than 10 µm was the major effect upon source atmospheric or dust cloud, as the windblown dust condition (Neff et al., 2013).Besides, other peaks were below 5 µm particle that was also suspended near ground obviously.However, the majority Aeolian dust size of transmission in the atmosphere was between 0.1 and 10 µm (Maring et al., 2003), due to the differences in the gravitational settling rate of each particle size (Mahowald et al., 2014).The result in this case revealed that the south site was evidently under the influence of Aeolian dust from the riverbed, especially in coarse size particle.
Table 2 exposed the vertical sampling data which included PM concentration, size distribution, and wind speed in each sampling height in this case.It has clearly demonstrated that Aeolian dust total mass concentration was increased, but the particle mass mode size was decreased, as the vertical sampling height increasing except one meter height.Besides, the particle deposition also made the most part of coarse particle remaining in the vicinage of the source.However, the Aeolian dust transportation distance at faster wind speed condition was farther than slower wind speed condition.According to the wind speed data and wind profile equation, the calculation of surface roughness height and friction velocity was about 1.07 mm and 0.68 m s -1 , respectively, at faster wind speed situation.

Comparison and Discussion
Comparing with these two cases, Fig. 7 showed the total particle mass concentration of each sampling sites.The tendency of vertical total particle mass concentration displayed the cross curves in the source site.In case I, the total particle mass concentration was decreased as increasing the sampling height due to the less fine particle suspension and the shorter coarse particle deposition distance under the slower wind speed condition.Contrarily, in case II, the total particle mass concentration was increased as increasing the sampling height except one meter height sampling under the faster wind speed condition.Moreover, as the wind speed  increased, the accumulation concentration of suspension particle was increased.Consistent with Hinds (1999), the faster the wind speed is, the greater the probability of reentrainment is.
In principle, according to the wind gradient, the parallel wind speeds near the ground approach to zero due to the non-slip condition.And the faster the wind is, the shorter the surface roughness height is (Hartmann, 2015).Due to the faster wind speed situation, the mass concentration of one meter height sampling was almost higher than other sampling heights.Hence, as presented in Table 2, the particle mode size of one meter height sampling was about 21.3 µm (d 50 by mass).In addition, this phenomenon was also happened under the slower wind speed condition as shown in the Table 1, however, the suspension particle was became smaller (14.8 µm, d 50 by mass).The results conformed to the physical theory of particle motion near the surface.
The result occurred by two processes, including erosion and denudation, especially in erosion process.Due to the wind active, the sand or gravel motion can make the corrasion and the attrition function.These are important processes of mechanical erosion in the earth's surface which may produce the smaller particulate.Besides, the particle size made the force difference, the larger the particle is, the lower the inter-particle force is (Kok et al., 2012).On the other hand, smaller particle might be hidden by the larger particle at the same flat that obstructed the fine particle resuspension.It was called the shielding effect of particle motion (Agudo et al., 2014;Wang and Arson, 2016).Therefore, the fraction of fine particle suspension was lower than coarse particle during the slower wind speed condition.However, if it's under the high wind force situation, the fine particle could suspend easier.As the most of health risk studies, the influence of fine Aeolian dust should be paid more attention (Kavouras et al., 2015;He et al., 2016).

CONCLUSIONS
The Aeolian dust from the semi-arid riverbed of Jhuoshuei River is an important PM emission source in Taiwan.According to the UAV photography and computational chromatic analysis, the percentage of riverbed covering conditions, including bare zone, wetlands, green covering and water covering was 50.1, 15.7, 16.8 and 17.4%, respectively, in January 5 th , 2011.However, in bare zone, partial land was covered with dry straws or dry sediment crust.The result Therefore, two different wind speed conditions were investigated and discussed in this study, the first case happened in November 26 th , 2010, and the second case was on January 15 th , 2011.The influence factor of downstream village's ambient air quality on river dust was the wind speed and distance between source and receptor.According to the horizontal results, the total mass concentration ratio of south site to source site (one meter height) was about 1:9 in first case and about 1:3 in second case.The ratio shown the influence index of Aeolian dust.The receptor was evidently under the influence of Aeolian dust from the riverbed, especially in coarse size particle in the second case.Moreover, the particle mass size characteristics of Aeolian dust were distributed in a bimodal or a multimodal curve, and the main peak size was above 10 µm.As the force balance between gravity settling and wind blowing of particle, the particle mode size (one meter height sampling) of faster wind speed case (about 21.3 µm, d 50 by mass) was larger than slower wind speed case (about 14.8 µm, d 50 by mass).Besides, in the first case, the total PM concentration was almost decreased with increasing sampling height, however, the second case was just the reverse.Shielding effect of particle motion made that the suspension of fine particle was more difficult than coarse particle during the slower wind speed condition.
According to the wind profile, the calculation surface roughness height of the case I and case II was about 3.08 mm and 1.07 mm, respectively, and the friction velocity was about 0.33 m s -1 and 0.68 m s -1 , respectively.Comparing with those two cases, the faster the wind is, the shorter the surface roughness height is.Owing to the above reason and the comparative result of those two cases, the faster wind speed caused the greater fine particle suspension and the higher suspension height.
Aeolian dust is an important issue, where covered not only in local but also in global.Besides, the phenomena of Aeolian dust combine lots of complicated processes.Consequently, this study clued the spatial variability of river dust events, which can further aid the site investigating, forecasting and preventing of dust influence.

Fig. 1 .
Fig. 1.Map shows the locations of sampling sites in study zone of target arid riverbed.Right part denotes the Jhuoshuei River within the index map of west-central Taiwan.

Fig. 2 .
Fig. 2. The unmanned aerial vehicle (UAV) imagery of target arid riverbed, besides, the photos under the UAV imagery shown two types crusting, water zone, wet lands, and dry straws covering on January 5 th , 2011.

Fig. 3 .
Fig. 3.The wind rose at the source site on November 26 th , 2010.

Fig. 4 .
Fig. 4. The PM mass concentration and size distribution of horizontal sampling by using model 298 Marple samplers on November 26 th , 2010.

Fig. 5 .
Fig. 5.The wind rose at the source site on January 15 th , 2011.

Fig. 6 .
Fig. 6.The PM mass concentration and size distribution of horizontal sampling by using model 298 Marple samplers on January 15 th , 2011.

Fig. 7 .
Fig. 7.The total PM mass concentration of two cases sampling by using model 298 Marple samplers.

Table 1 .
The vertical sampling PM properties and wind speeds of "case I" in each sampling height.

Table 2 .
The vertical sampling PM properties and wind speeds of "case II" in each sampling height.