Effects of Particulate Matter of Various Sizes Derived from Suburban Farmland , Woodland and Grassland on Air Quality of the Central District in Tianjin , China

Poor air quality directly affects human health and has become an increasingly important environmental issue in Tianjin, China. The suspension of particulate matter (PM) in the atmosphere is not only from industrial pollutants, but also soil wind erosion; however, the contributions from farmland, woodland, and grassland have rarely been considered in this region. We conducted an assessment of PM sources through wind erosion, dust emission, and dust transportation from urban and rural areas to the central district in Tianjin, and our results demonstrated that the spatial variability of wind erosion and dust emission strongly depends on land use, particle size distribution and meteorological conditions. The equations in this study were empirical, and soil properties such as aggregation and crusting, as well as surface characteristics such as canopy height and residue cover, were not considered. The dust emission capacity of woodland and grassland was the lowest because of vegetation coverage. The values obtained in this study may overestimate emissions, because soil aggregation was not considered. The yearly dust amounts of PM15–20 (particles with aerodynamic diameter from 15 μm to 20 μm), PM10–15 (particles with aerodynamic diameter from 10 μm to 15 μm), and PM10 (particles with aerodynamic diameter less than 10 μm) from wind erosion in 2009 from the urban area in Tianjin were estimated as 5,400 t, 5700 t and 17,300 t, respectively, while those from the rural area were 14,000 t, 15,300 t and 40,700 t, respectively. The dust emission contributed from farmland accounted for 99.5%, and that from woodland and grassland only accounted for 0.5%. The PM10 transported to the central district and PM10 concentrations in the days with the 20% highest PM10 concentrations in the central district in 2009 were compared. The R was 0.74, which meant the two variables were highly correlated.


INTRODUCTION
Poor air quality has become a serious environmental problem in Tianjin, China.Particulate matter with an aerodynamic diameter of less than 10µm (PM 10 ) is a major air pollutant in Tianjin.PM 10 was the principal pollutant in 71.5% of the days (261 out of 365 days) in 2009 (TBEP, 2010).Dust Storms frequently occurred in Tianjin, especially during the sand period, and therefore the coarser particulate matters such as PM 10-15 and PM 15-20 were also of our concern.Particulate matter in the atmosphere has been noted for its effects on visibility (Hyslop et al., 2009) and global climate change, primarily a cooling effect due to increased scattering light to space as the atmospheric aerosol burden increased (Chen et al., 2003), as well as on human health.Particulate matter adversely affects the respiratory system and contributes to lung cancer, pulmonary disease, and asthma (Dockery et al., 1993).
Particulate matter is tiny solid matter suspended in air; it originates not only from a variety of mobile and stationary sources (automobiles, power plants, industrial processes, etc.), but also from wind erosion of agricultural land.Wind erosion on agricultural land is a serious environmental problem in northern China, especially in the arid and semiarid regions.Wind erosion can change surface soil properties, degrade soil fertility, and, more importantly, degrade air quality.Fine particulates eroded from farmland and subsequently suspended in the atmosphere recently were raised concern about air quality in the western United States (Sharratt et al., 2007).Saxton (1995) reported that windblown dust from farmland was the major noncompliance source of the U.S. EPA National Ambient Air Quality Standard for PM 10 within the Columbia Plateau region in the Pacific Northwest.Dust from soil erosion is also the major source of particulate matter in the atmosphere in Tianjin, China, most notably during the dry season (from October to May).Recent observation of air quality in Tianjin showed that PM 10 concentration was higher in October, November, and December than in other months in 2009 (TBEP, 2010).The sources of PM 10 in Tianjin were analyzed using a chemical mass balance (CMB) receptor model, and the PM 10 contribution of soil dust source was 104 µg/m 3 (39%) in 2002 (Bi et al., 2007).
Accurately estimating soil loss from wind erosion is essential for conservation planning, natural resource inventories, and air quality monitoring.Simulation of soil loss from wind erosion is a difficult process as emission capacity is affected by many factors including wind speed, soil texture, and land use (Goossens et al., 2011;Juneng et al., 2011;Sharratt et al., 2011).
Wind erosion is a complex, dynamic process consisting of detachment, transport, and deposition of soil particles (Skidmore et al., 1986;Zobeck et al., 2006).Soil particulates normally move in three ways in the wind erosion process: creep, saltation and suspension.Soil erosion by wind is initiated when the wind speed exceeds the saltation threshold velocity for a given soil at a surface condition.Most of the eroding material moves only a short distance above the soil surface by creep and saltation of larger particles (Grivas et al., 2008;Zhang et al., 2011).The bouncing and abrasive action of larger particles on the soil surface gives rise to suspension of small particulates that can be carried over thousands of kilometers in the atmosphere (Karaca et al., 2009;Martet et al., 2009).Models have been developed in recent years to simulate soil erosion and dust emission to the atmosphere (Holmén et al., 2001;Zender et al., 2003;Funk et al., 2004;Vautard et al., 2005;Visser et al., 2005;Shaw et al., 2008;Kavouras et al., 2009), but the quantity of the dust moving from the urban area to the central district in the air has not been thoroughly estimated.
The main purpose of this study was to estimate dust emissions from farmland, woodland and grassland, and to evaluate their impact on air quality.The objectives of the research included: (1) to estimate the wind erosion modulus and the dust emission modulus and the dust emission amount from each district/county in the urban and rural areas of Tianjin; (2) to compare the effects of wind erosion and dust emission on pollution from farmland, woodland, and grassland in the central district; and (3) to assess the transport of PM 10 , PM 10-15 and PM 15-20 from urban and rural areas to the central district.
Tianjin is located in the northeastern part of the Huabei Plain adjacent to the Bohai gulf.It receives the typical continental monsoon climate: cold and dry in winter and hot and wet in summer.The average annual temperature is around 12.3°C, and the annual precipitation ranges from 550 to 680 mm.About 82% of the total annual precipitation is distributed in the summer and fall.Winds are predominantly from the northwest in the winter and southeast in the summer.Strong winds always occur in the spring and are accompanied by dust storms.Tianjin is an industrial city, with major industries ranging from electronics, automobiles, petrochemicals, metallurgy, biomedical industry to alternative energy.
The lands of the urban and rural areas in Tianjin are divided into potentially erodible lands including farmland, woodland, grassland, bare land, and non-erodible construction land or land covered by water.The land use type dataset was obtained from Tianjin Bureau of Environmental Protection (TBEP).The land use map of Tianjin was shown in Fig. 2. The construction land, including buildings, streets and commercial land, was considered to contribute limited amounts of blown dust (TBEP, 2010).So both water-covered and construction lands were regarded as non-erodible and consequently not included in our analysis.

Sample Collection and Analysis
Based on soil profile characteristics, soil types, and land use, we collected in the summer of 2009 a total of 398 soil samples from erodible lands, of which 343 are from farmland, 39 from woodland, and 16 from grassland (see Fig. 3).Among these samples, 94 are taken from the urban area and 304 from the rural area.The soil samples were collected from a depth of 0-15 cm.Obvious extraneous matter was not collected in sampling.The coordinates of sampling locations were recorded with a GPS.
Soil samples were dried in the laboratory and screened by a 60-mesh nylon sieve.Particle size distribution of all samples was analyzed with a BT-9300S Laser Particle Size Analyzer.The instrument has a capability of measuring particle sizes from 0.1 to 340 µm.Thus, three particle sizes (PM 10 , PM 10-15 , and PM 15-20 ) of surface soil were measured.The repeatability error of this instrument is less than 1%.
The wind data including wind speed and direction were collected at the weather station of each district/county in Tianjin.The wind speed threshold, at which particle motion just begins, is a critical parameter of any accurate formulation that expresses the transport rate of particles such as sand and soil as a function of horizontal wind-speed in the Earth's atmosphere.The wind threshold velocity was set as 5 m/s in the urban and rural areas according to the research on soils in northern China (Chen et al., 1994) and wind tunnel experiments (Zhang et al., 2005).The height of the wind speed threshold of 5 m/s in this study was assumed to be 10 meters.

Wind Erosion and Dust Emission
To estimate wind erosion and dust emission, we used the following empirical formulas based on wind speed and surface soil types.The formulas were developed and validated through wind tunnel experiments with soils in Northwestern China (Gao et al., 2008).The formulas were limited for soils from farmland, woodland, and grassland, especially in an arid and semi-arid region.And so soil water content was not regarded as a main factor to affect wind erosion and dust emission and was not considered in their formulas (Zhang et al., 2005).

Wind Erosion Modulus
Wind erosion modulus refers to the average amount of wind erosion dust in per unit of time and per unit of area.The equations in this study were empirical, and soil properties such as aggregation, crusting and surface characteristics where Q fa is a wind erosion modulus on farmland t/(ha).C  is a scale-revised coefficient of about 0.0018; A is the wind speed revised coefficient of about 0.89; U j is the j th hourly wind speed.Z 0 is surface aerodynamic roughness (cm), about 0.55 cm in northern China; T j is the accumulation time in which the wind speed is U j and the wind erosion occurred (min).The threshold of wind speed is defined as 5 m/s on dry farmland in northern China and is slightly higher on sandy surfaces.j is the wind speed category.The first category of wind speed is 5-6 m/s, and the average wind speed is 5.5 m/s, so U j=1 = 5.5 m/s, U j=2 = 6.5 m/s, and so on.The highest U j is the maximum wind speed recorded in the category at the weather station.The height of the wind speed observation is 10 meters.The estimation of wind erosion modulus on woodland and grassland was calculated with the following equation: 10 exp 2.4869 0.0014 54.9472 / ( ) where Q fgf is the wind erosion modulus of woodland or grassland whose coverage of vegetation is VC[t/(ha•yr)].U j is the jth hourly wind speed, which is higher than threshold value.T j is the accumulation time in which the wind speed is U j (min).C  is the scale-revised coefficient of about 0.0018; A is the wind speed revised coefficient, about 0.89.

Dust Emission Modulus
Dust emission modulus refers to the average amount of the wind erosion dust that actually enters the atmosphere in per unit of time and per unit of area.
The dust emission from farmland was estimated with the wind erosion modulus, using the following equation: where Q fd is farmland dust emission modulus and a is the mass percentage of particulates (PM 10 , PM 10-15 and PM 15-20 ) from farmland.Eq. ( 4) was used to estimate dust emission on woodland and grassland: where Q fgd is woodland and grassland dust emission modulus and b is the mass percentage of particulates(PM 10 , PM 10-15 and PM 15-20 ) from woodland and grassland.

Estimating Dust Emission Transported to the Central District
The wind speed and direction in the central district were recorded from the routine measurement at a height of 10 m.The starting wind speed for dust emission was set at 5 m/s with a 1-h interval.Wind direction frequencies were calculated for each of the 16 directions.
The 16 wind direction lines to the center of the central district were drawn.An irregular sector was formed based on the two lines on either side of the wind direction line ± 11.25° and the boundary of the city.The sector area of each wind direction was calculated.For example, the sector of the N direction contained the red, green, blue and yellow areas as shown in Fig. 3.As the farthest distance from the central district to the boundary of Tianjin is 128 km, and the transportation distance of dust with D = 20 µm is 84.7 km at the wind speed of 5 m/s, 101.7 km at 6 m/s, and 118.6 km at 7 m/s (Gao et al., 2008; see Table 1), the red area indicated the dust emission area when the wind speed was 5 m/s; when the wind speed increased to 6 m/s, the dust emission area extended to the green area until it extended to the yellow area at the greater wind speed.The same applied to the other wind directions.
The dust transportation capacity of wind speeds from the urban and the rural to the central district was calculated with the following formula: where λ i , I = 1,2,3, is the mass percentage of dust particles in the surface soil with D ≤ 10 µm, 10 < D ≤ 15 µm, and 15 < D ≤ 20 µm, respectively.S hjk is the sector area with the k wind direction and the U j , j = 5,6,7,…,17; wind speed.Q jk

Particle Size Composition of Surface Soil
We generated the spatial distribution map of topsoil particle size composition by kriging and classified the surface soil in the urban and rural areas into nine categories based on particle size composition with ArcGIS 9.2 (Fig. 4); the soil particle size distribution was listed in Table 2.The finest soil particle size distribution was found in DL, with D ≤ 10 µm up to 57% and D ≤ 20 µm up to 78%, and BC and JN with D ≤ 10 µm comprising 49% and 43% of the surface soil, respectively.These three districts all belong to the urban area.The maximum particle size distribution was found in JX, the farthest rural district north of the central district, with D ≤ 10 µm comprising 18% and D ≤ 20 µm comprising 35% of the surface soil.In general, the soil particle size distribution is coarser in the northern and the western regions than in the southern and the eastern regions of Tianjin.The distribution of soil particle size distribution is consistent with the topography in Tianjin, which is high in the northwest and low in the southeast.

Wind Regime
The wind regime in 2009 was collected at one station in each district of Tianjin including hourly wind speeds and directions.The accumulation time of wind speed in each category was calculated and listed in Table 3.The highest wind frequency with wind speed over 5 m/s was found in JN with over 900 h, followed by WQ, TG, HG, and XQ, with 743, 728, 713, and 695 h, respectively.The lowest frequency was located in JX with 112 h, followed by BC and JH with 211 and 273 h, respectively.The highest wind speeds recorded in 2009 were found in WQ with speeds at 15-16 m/s for 2 h and at 16-17 m/s for 1 h.Thus, the strongest wind over the threshold occurred in the northwest and the southeast regions.

Wind Erosion Modulus
Wind erosion was not supposed to occur during the rainy season or when the ground was covered with snow.We ran the simulation of wind erosion with soil in a dry condition.
The rainy and snowy days were removed from the simulation.The wind erosion modulus of each district was calculated and listed in Table 4.The total soil wind erosion from the urban and rural areas in 2009 was estimated as 173,200 t, including 171,600 t from farmland, which was over 99% of the total.The total erosion from woodland and grassland was only estimated as 1,700 t, less than 1% of the total.The yearly wind erosion modulus was about 13 t/ha from farmland and 5 t/ha from woodland and grassland in 2009.
The highest erosion modulus was found in JN with 27 t/ha, followed by WQ with 24 t/ha from farmland.Both JN and WQ were found with the highest frequency of wind speeds over 5 m/s.The average modulus from farmland in the four urban districts was 14 t/ha compared with the average modulus in the rural area of 12 t/ha from farmland.This means that the wind erodibility was higher in the urban area than in the rural area.

Dust Emission
Dust emission modulus was calculated with the dust emission equation in each district and the results were listed in Table 5.Total dust emission in 2009 was estimated in the urban and rural areas up to 98,500 t for particle diameter size (D) less than 20 µm, including 58,000 t (D ≤ 10 µm), 21,000 t (10 < D ≤ 15 µm), and 19,400 t (15 < D ≤ 20 µm).PM 10 was the dominant component of dust emission, which took up to 60% of the total particle size composition and was the major source of dust in the central district.PM 10-15 and PM 15-20 were 21% and 19% of the total dust emission, respectively.
The dust emission modulus in the urban area was obviously higher than in the rural area.The dust emission modulus on farmland in the urban area was 6 t/ha with D ≤ 10 µm, 1.7 t/ha with 10 < D ≤ 15 µm, and 1.5 t/ha with 15 < D ≤ 20 µm; in contrast, dust emission modulus on farmland in the rural area was 4.1 t/ha with D ≤ 10 µm, 1.4 t/ha with 10 < D ≤ 15 µm, and 1.3 t/ha with 15 < D ≤ 20 µm.The average dust emission modulus on woodland and grassland in the urban area was 1.1 t/ha with D ≤ 10 µm, 0.3 t/ha with 10 < D ≤ 15 µm, and 0.3 t/ha with 15 < D ≤ 20 µm, but the average dust emission modulus in the rural area was 0.6 t/ha with D ≤ 10 µm, 0.2 t/ha with 10 < D ≤ 15 µm, and 0.2 t/ha with 15 < D ≤ 20 µm.The regional difference was mainly determined by the particle size composition.The fine soil particle size distribution in the urban area provided more dust sources for emission.Because of the increase in the surface coverage of woods and grasses, the dust emission from woodland and grassland was much lower than that from farmland.The dust emissions in both the urban and rural areas were mainly contributed from farmland, with over 99.45% of the total emission from all the three types of land.

Dust Estimation from the URBAN and RURAL Areas to the Central District
Based on the weather station data, we generated the wind frequencies of each district in the urban and rural areas with the wind direction pointing to the central district in 2009 as shown in Fig. 3 and listed in Table 6.For example, with winds coming from the north direction (N) or the northnortheast (NNE), the districts that may bring about potential dust emission to the central district include JX, BD, WQ, NH, BC, and DL.
We estimated the dust of various particle sizes blown from the urban and rural areas to the central district (Table 6).In 2009, PM 20 from the urban and rural areas to the central district were 45,800 t, including PM 10 of 26,900 t, PM 10-15 of 9,800 t, and PM 15-20 of 9,000 t.In the urban area, PM 20 was 11,300 t, including PM 10 of 6,800 t, PM 10-15 of 2,300 t, and PM 15-20 of 2,300 t.In the rural area, PM 20 was 29,900 t, including PM 10 of 16,900 t, PM 10-15 of 6,700 t, and PM 15-20 of 6,300 t.
Most of the PM 10 transported to the CD was contributed from the N or NNE direction.The PM 10 contribution was up to 30.57% from WQ, and up to 15% was from BD and NH.Minimal contribution was from DG, HG, and JX with less than 1%.In general, PM 10 transported to the central district was mostly contributed from the rural area with 71.31%;only 28.69% came from the urban area.
PM 20 from the urban and rural areas to CD was 45,800 t, including PM 10 of 26,900 t.Windblown dust originated from the urban and rural areas, especially from farmland, is the primary source of PM 10 in the central district, although there are other sources such as unpaved roads and construction lands.The estimation of PM 10 from farmland may be overestimated due to the use of particle size distribution.We did not consider the effect of soil aggregation structure.

Comparison of PM 10 Transported with Concentration in CD
The days with the 20% highest PM 10 concentrations in 2009, which were from 0.135 mg/m 3 to 0.415 mg/m 3 , were selected, and the PM 10 transported to CD during these days were estimated based on the above-mentioned method.The correlation between the PM 10 transported to CD and PM 10 concentration in the days with the 20% highest PM 10 concentrations in CD in 2009 was shown in Fig. 5.The R 2 was 0.74, which meant the two variables were highly correlated.

CONCLUSIONS
Windblown dust emission was treated as a non-continuous spatio-temporal process (Korcz et al., 2009).The spatial distribution of wind erosion and dust emission was strongly determined by wind speed over threshold velocity.Winds with northwest and southeast directions are the dominant wind regime in Tianjin.The northwest winds prevailing in the winter and the spring are the main forces that cause dust emission.In the rural area, the northern districts excluding JX had higher wind speeds over the threshold and provided more dust emission to the central district compared with the southern districts (Fig. 6).In the urban area, XQ was the main contributor of dust emission to CD compared with the other three districts.
DG and HG contributed minor amounts of dust emission to CD.The prevailing wind directions with wind speed > 5 m/s were NNW, N, and E in DG and NNW and NW in HG; thus, dust emission from these two districts were rarely blown into CD.
There are some limitations in this study.It only accounted for particle size distribution, not aggregate size distribution though it may also affect wind erosion.Therefore, the values obtained in this study may overestimate emissions.Besides, the equations in this study were empirical, and soil properties such as aggregation, crusting and surface characteristics such as canopy height, residue cover were   not considered.In this study, soil samples were screened by a 60-mesh nylon sieve, and the soil aggregation with the diameter more than 0.3 mm was excluded from estimating the dust emission.Particle size distributions of the 398 samples were analyzed using a Laser Analyzer, but the water used may destroy the soil aggregation structure with the diameter less than 0.3 mm.We did not consider soil water content in the simulation, although the factor is important for wind erosion.We supposed the soils were in a drought condition with low water content and the water content had minor influence on surface soil emission.Actually, soil water content was very low in summer and spring in Northern China.Dust transportation is also regarded as a complex procession.The particulate matter may have creep, saltation, and suspension depending on the carriage of wind energy.The simulation of our research was only to estimate the potential contribution of particulate matter from farmland in Tianjin.Over 99% of all dust emission was contributed by farmland, only a small amount from woodland and grassland.Most dust emission came from the four districts of WQ, XQ, BD and NH, which were located to the north or northwest of CD, with north as their prevailing wind direction.
The four urban districts have fine particle size distribution with dominant percentage of fine particulate matter, especially the composition of PM 10 .The fine particle size distribution soils in the urban area provide the potential sources of suspension to CD.Although PM 10 may come from other sources such as constructed lands, unpaved roads, and automobiles, the farmland in the urban area of Tianjin were still regarded as the main source of dust emission, comprising up to 60% of the total emission.
Through this research, we found that up to 4.7 t/ha of PM 10 was lost from farmland in 2009.The emission of PM 10 was much higher than that in the U.S. Pacific Northwest, which was only 0.4 t/ha annual loss according to a wind erosion processing system (WEPS) simulation in the summer fallow land (Feng and Sharratt, 2007).Because no field observation and field experiments of dust emission were conducted, it is difficult to validate the dust emission.The model simulation results may be overestimated as we considered the whole soil composition to suffer wind erosion, without considering the surface condition.In future study, soil surface emission can be measured with wind tunnel experiments and WEPS simulations with the support of the local climatic data and land use to further compare with the results from this research.

Fig. 3 .
Fig. 3. Dust emission areas under the condition of 16 wind directions at different wind speeds (5 m/s-18 m/s).
is the wind erosion modulus of the h district (BC, XQ, JN, DL, JX, JH, TG, HG, DG, BD, NH, WQ) with the U j wind speed.The dust with D ≤ 15 µm would be transported for a long distance with surpassing the administrative boundary if the dust were released into the air; so the dust emission area is the sector area with different wind directions.Q fdc is the farmland dust estimation from the urban and the rural to central district, and Q fgdc is the woodland and grassland dust estimation from the urban and the rural to central district.All work of this study was based on the data collected in 2009.

Fig. 4 .
Fig. 4. Particle size composition of surface soil in the urban and rural areas in Tianjin, China, based on Kriging interpolation.

Fig. 5 .
Fig. 5. Comparison of PM 10 transported to the central district and PM 10 concentrations in the days with the 20% highest PM 10 concentrations in 2009.

Fig. 6 .
Fig. 6.Wind erosion flux, dust emission flux, PM 20 transported to the central district, and PM 10 transported to the central district of each district in the urban and rural areas

Table 1 .
The transportation distance (km) of the dust with different particle sizes (D = 20 µm, 15 µm and 10 µm) under the various wind speeds.

Table 2 .
Particle size composition of surface soil in 12 districts in Tianjin, China (%).

Table 3
The accumulation time of wind speed at different levels from weather stations in 2009 (h).

Table 4 .
Estimated wind erosion modulus and wind erosion amount from each district.

Table 5 .
Estimated dust emission modulus and dust emission amount of each district.

Table 6 .
The dust of various sizes transported from the urban and rural areas to the central district in 2009 (t).