Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Modeling and Analysis of Source Contribution of PM10 during Severe Pollution Events in Southern

W. C. Wang, K. S. Chen*

Institute of Environmental Engineering, National Sun Yat-Sen University, 80424, Taiwan, ROC.

Abstract

This work simulates the hourly variations of PM10 (suspended particles with diameter < 10 μm) during severe pollution events in southern Taiwan (Kaohsiung City and Pingtung ) in spring, autumn and winter of 2005 by using the Air Pollution Model (TAPM). Comparisons between simulations and measurements at three sites (industrial, urban and rural) were satisfactory. The synoptic weather chart indicated that prevailing winds were northwest (spring), north (autumn), and northeasterly (winter). Meteorological conditions suggest that PM10 typically accumulated and triggered a pollution episode on days with high surface pressure and low winds. Estimations using the TAPM model suggest that point-source emissions were the predominant contributors (about 49.1%) to

PM10 concentrations at Hsiung-Kong site industrial site in Kaohsiung City, followed by area sources (approximately 35.0%) and transport from neighboring areas (7.8%). Because (urban) and Chao-Chou town (rural) are located downwind of Kaohsiung City when north or northeasterly winds prevail, the two sites also experience severe pollution events despite the lack of industrial sources; transport from neighboring areas contributed roughly 39.1% to PM10 concentrations at Pingtung site and 48.7% at Chao-Chou site. Since traffic emissions contributed little (around 8%) to

PM10 concentrations at the three sites, reducing PM10 emissions from industrial sources in Kaohsiung City should be an effective way of improving air quality for Kaohsiung City and downwind areas such as .

Keywords: TAPM; Particulate matter; PM10; Air pollution modeling; Source contribution; Meteorological condition.

INTRODUCTION

* Corresponding author. Tel.: +886-7-5254406; Fax: Kaohsiung City, in southern Taiwan, is a +886-7-5254406 heavily industrialized harbor city, with an E-mail address: [email protected] area of 153.6 km2 and a population of 1.49

319 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

2560000

2550000 Taiwan D C Strait 2540000

2530000

2520000

2510000 B A 2 2500000 UTM-N (m) UTM-N UTM-E (m) N 2490000 1 3

2480000

2470000

160000 170000 180000 190000 200000 210000 220000 230000 240000 250000 Pacific UTM-E (m) Ocean

A: Kaohsiung City B: Pingtung County C: D: County Monitoring sites (1: Hsiung-Kong, 2: Pingtung, 3: Chao-Chou) Industrial Parks

Fig. 1. Model domain and the three monitoring sites in southern Taiwan. million. Six major industrial parks are industries located in northern Pingtung located in and around Kaohsiung City (Fig. County. However, air quality in northern 1). These parks are home to oil refineries, (e.g., Pingtung City) and central (e.g., petrol/plastic industries, power plants, Chao-Chou town) Pingtung County is as bad iron/steel/metal plants, recycling factories, as that in the Kaohsiung area, despite the and large municipal waste incinerators. Due fact that population densities, traffic to intensive industrial and traffic activities, volumes and industrial emissions in the Kaohsiung metropolitan area has the Pingtung County are significantly lower poorest air quality in Taiwan − either than those of the Kaohsiung area. This poor increased ground-level concentrations of air quality is mainly because northern and particulate matter (PM) or ozone (O3) central Pingtung County are downwind of associated with unfavorable meteorological the Kaohsiung area when north or conditions − particularly between late fall northeasterly winds prevail typically in and mid-spring (Chen et al., 2004). Pingtung autumn and winter (Chen et al., 2003, County, to the south of Kaohsiung City, is 2004). primarily agricultural with several small Primary PM is emitted directly into the

320 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

atmosphere by anthropogenic sources (e.g., MODEL DOMAIN AND industry, vehicles, combustion sources, MONITORING SITES bare lands and open burnings), and natural sources (e.g., volcanic eruptions, wildfires The three monitoring sites were and marine aerosols). Atmospheric PM can Hsiung-Kong (industrial), Pingtung (urban) and carry acids and toxic species (e.g., Chao-Chou (rural) sites; all are located in polycyclic aromatic hydrocarbons and southern Taiwan (Fig. 1). Taiwan’s heavy metals) and can have adverse effects Environmental Protection Administration (EPA) on human health (Cheng et al., 1996). has set up an air-quality monitoring station at Epidemiological studies have demonstrated each site that collect hourly air quality and

a strong relationship between elevated meteorological data, including that for PM10

concentrations of PM10 and mortality and concentration, temperature and wind. morbidity (Lin and Lee, 2004; Arditsoglou Hsiung-Kong, with a population of and Samara, 2005). Identification of air approximately 151,932 and an area of 39.9 km2, pollution sources is important in is adjacent to the Linhei Industrial Park at the developing clean-air strategies (So and southern end of Kaohsiung City. This industrial Wang, 2004; Wang and Shooter, 2004; park is home to iron/steel industries, Viana et al., 2006). In non-attainment areas, petrol/chemical plants, power plants, secondary air-pollution models are frequently aluminum sinter ovens and a large municipal employed to predict hourly pollution levels waste incinerator. Since Kaohsiung and thereby help establish cost-effective International Airport is located in the means of reducing atmospheric PM Hsiung-Kong , traffic density is usually concentrations and controlling emission very high. sources (Park et al., 2004; Zawar-Reza et Pingtung, which has a population of roughly al., 2005; Luhar et al., 2006; Wilson and 216,222 and an area of 65.1 km2, is the capital Zawar-Reza, 2006; Cheng et al., 2007). of Pingtung County. Pingtung is primarily an

In this work, hourly PM10 variations were administrative and business city. Chao-Chou simulated using TAPM-3.0 (Hurley, 2005) has a population of about 57,189 and covers for severe pollution events during the three 42.4 km2. It is primarily an agricultural rural worst seasons (spring, fall, and winter) in town located in Pingtung County. Traffic Kaohsiung City and Pingtung County. Each density is generally low. The distance between simulation covers three consecutive days or the Hsiung-Kong and Pingtung monitoring sites 72 h. Measured data from three monitoring is about 19 km, that between Pingtung and sites were collected and compared with Chao-Chou sites is about 16 km, and that simulation results. Meteorological between Hsiung-Kong and Chao-Chou sites is conditions and source contributions at each about 23 km. monitoring site were analyzed.

321 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

TAPM MODEL horizontal diffusion coefficient; w′φ′ is the

eddy term; f is the Coriolis parameter; θv is Governing equations and grid setup the potential virtual temperature. The The Air Pollution Model (TAPM) is a turbulence terms in Eqs. (1) and (2) are three-dimensional, prognostic, Eulerian, derived by solving equations for the incompressible, non-hydrostatic, primitive turbulent kinetic energy and the eddy equation model in terrain-following dissipation rate. The governing equation for coordinates for simulating atmospheric species concentration, χ, such as PM10 is motion and pollutant transport using nested grids. Hurley et al. (2003) and Hurley (2005) dχ ∂ ⎛ ∂χ ⎞ ∂ ⎛ ∂χ ⎞ ⎜ ⎟ ⎜ ⎟ described in detail the governing equations = ⎜ K χ ⎟ + ⎜ K χ ⎟ dt ∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ for mass, momentum, and energy, which are χ ⎛ ∂σ ⎞ ∂ briefly elucidated as follows. − ⎜ ⎟ ()w′χ′ + R + S (5) ⎜ ∂z ⎟ ∂σ x ⎝ ⎠ d u ∂ u ∂ ⎛ ∂ u ⎞ = + (u,v) ⋅ ∇ = ⎜ K ⎟ + ⎜ H ⎟ where K is the diffusion coefficient; d t ∂ t ∂ x ⎝ ∂ x ⎠ χ w′χ′ is the eddy term; R is the chemical ∂ ⎛ ∂u ⎞ ∂w′u′ ∂σ ⎛∂P ∂P ∂σ ⎞ χ ⎜K ⎟− −θ ⎜ + ⎟+ ⎜ H ⎟ v⎜ ⎟ reaction term; S is the pollutant emission ∂ y⎝ ∂ y⎠ ∂σ ∂z ⎝ ∂x ∂σ ∂x ⎠ χ term. The diffusion coefficient utilized for f v − N s (u − us ) (1) pollutant concentration is K χ = 2.5 K, d v ∂ v ∂ ⎛ ∂ v ⎞ where K is the diffusion coefficient for the ⎜ ⎟ = + (u, v) ⋅ ∇ = ⎜ K H ⎟ d t ∂ t ∂ x ⎝ ∂ x ⎠ turbulent kinetic energy. Furthermore, 13 ∂ ⎛ ∂v ⎞ ∂w′v′ ∂σ ⎛∂P ∂ P ∂σ ⎞ species are involved in ten reactions. These ⎜ ⎟ ⎜ ⎟ + ⎜KH ⎟− −θv⎜ + ⎟ ∂ y ⎝ ∂ y⎠ ∂σ ∂ z ⎝ ∂ y ∂σ ∂ y ⎠ species are NO, NO2, O3, SO2, hydrogen smog − f u − N s (v − vs ) peroxide (H2O2), R , radical pool (RP), (2) stable non-gaseous organic carbon (SNGOC), stable gaseous nitrogen (SGN) ∂σ ⎛ ∂ u ∂ v ⎞ ∂ ⎛ ∂σ ⎞ ∂ ⎛ ∂σ ⎞ & = −⎜ + ⎟ + u ⎜ ⎟ + v ⎜ ⎟ products, stable non-gaseous nitrogen ∂σ ⎜ ∂ x ∂ y ⎟ ∂σ ⎜ ∂ x ⎟ ∂σ ⎜ ∂ y ⎟ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ (SNGN) products, stable nongaseous sulfur (3) (SNGS)products, Airborne Particulate ⎛ z − z ⎞ Matter (APM), and Fine Particulate Matter σ = z ⎜ S ⎟ (4) T ⎜ ⎟ (FPM), which comprises secondary ⎝ zT − zs ⎠ particulate concentrations of SNGOC, In the above, t is time; (u, v) are SNGN and SNGS (Hurley, 2002). In this horizontal winds; us and vs are large-scale work, pollutant emission rates were synoptic winds; σ is sigma-pressure;σ& is considered and input to the model via vertical wind; zT is the height of the top of boundary conditions (discussed later). the model; zs is terrain height; KH is the The TAPM is run on a personal computer

322 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

for simulating mesoscale atmospheric multiplicative factor of 0.0067 was used for motions using nested grids with converting the emission rates of NMHC to meteorological, geographical and air those of Rsmog (Johnson, 1984), and a pollution components. Each horizontal layer multiplicative factor of 1.0 was applied to had 35 × 35 grids, nested from the outside to emission rates of all other primary pollutants, the inside with size of 20, 7.5, 2 and 0.5 km. such as PM10, NOx, and SO2. The entire simulation domain covered over TAPM classifies the surface vegetation 700 km × 700 km (Fig. 1); the fine grids (land-use) into 29 classes. Surface data in covered the southern Taiwan over the the model were acquired using geographical domain of 262.5 km × 262.5 km, and were charts that were issued by the Ministry of centered at 120o22’ E and 22o23’ N. The the Interior, Taiwan, to determine area vertical domain was 8000 m high and fractions of land-use in each grid. comprised 25 horizontal layers at altitudes The TAPM model is initialized at each of 10, 25, 50, 100,150, 200, 250, 300, 400, grid point with values for wind velocities, 500, 600, 750, 1000, 1250, 1500, 1750, 2000, temperatures and humidity ratios that were 2500, 3000, 3500, 4000, 5000, 6000, 7000 interpolated from synoptic analyses. The and 8000 m. initial pollutant concentrations at inflow boundaries on the outermost grids are set to Emission inventory and boundary background values, while zero-gradient conditions conditions are applied at outflow boundaries. An emission inventory was obtained using Background conditions for pollutants were 3 TEDS-6.03 (2006), which was issued by set to 25 μg/m for PM10, 0.7 ppbv for Rsmog

Taiwan’s EPA. Table 1 presents all emission (Hurley, 2003), 1 ppbv for SO2 and 3 ppbv rates of PM10, sulfur dioxides (SO2), for NO2 (Chang and Cardelino, 2000). nitrogen oxides (= NO + NO2) and Notably, the initial values of eddy terms non-methane hydrocarbons (NMHC) from were set to zero because of a thermally point (industrial), area and line (traffic) stable condition at midnight. sources in each region. The methods of Four-dimensional data assimilation (FDDA) obtaining emission rates from various was adopted to compare simulated surface sources, except the updated emission rates in winds with ground observations and correct TEDS-6.03, were similar to those used by horizontal momentum equations using the TEDS-4.2 (Chen et al., 2003). However, Newtonian relaxation (or nudging) reactive hydrocarbons, Rsmog, replace the procedure (Stauffer and Seaman, 1994; NMHC rate in the TAPM (Johnson, 1984), Hurley, 2002). in which Rsmog is defined as the product of a Model performance was assessed relative reactive coefficient (or a multiplicative to actual measurements using the correlation factor) and NMHC emissions. A coefficient (R) and index of agreement

323 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Table 1. Emission rates of pollutants in TEDS-6.03 in southern Taiwan (kton per year)

PM10 SO2 NOx NMHC Paved Road 12.148 0.000 0.000 0.000 Area Sources Unpaved Road 0.000 0.000 0.000 0.000 Kaohsiung Others 8.061 4.579 11.169 47.198 City Gasoline Vehicles 1.002 0.061 4.512 17.761 Line Sources Diesel Vehicles 1.064 0.179 6.720 1.142 Point Sources 7.989 46.953 61.347 39.038 Paved Road 8.607 0.000 0.000 0.001 Area Sources Unpaved Road 1.713 0.000 0.000 0.020 Kaohsiung Others 6.639 0.735 2.502 55.413 County Gasoline Vehicles 0.761 0.046 3.973 11.907 Line Sources Diesel Vehicles 1.373 0.230 7.923 0.946 Point Sources 5.872 29.894 35.741 32.088 Paved Road 7.965 0.000 0.000 0.000 Area Sources Unpaved Road 1.367 0.000 0.000 0.016 Pingtung Others 9.561 0.787 6.380 27.255 County Gasoline Vehicles 0.483 0.034 2.790 7.536 Line Sources Diesel Vehicles 0.706 0.127 3.827 0.550 Point Sources 0.868 0.641 0.713 2.912 Paved Road 9.928 0.000 0.000 0.008 Area Sources Unpaved Road 1.481 0.000 0.000 0.046 Tainan Others 14.740 1.010 2.570 46.231 Area* Gasoline Vehicles 0.713 0.054 4.360 8.540 Line Sources Diesel Vehicles 1.580 0.267 8.469 1.019 Point Sources 4.659 10.437 6.744 35.028 * Tainan area includes Tainan City and Tainan County.

(IOA) as follows (Willmott et al., 1985). variations in pressure and wind speed for March 8–10, 2005. A high-pressure system N 2 ∑i=1()Pi − Oi and northwest winds prevailed in southern IOA = 1 − (6) 2 Taiwan (Fig. 2(a)). Pressure was highest at ∑N ()P − O + O − O i=1 i i approximately 1023.4 hPa on March 8 and where Pi and Oi are predicted and measured lowest at about 1010.8 hPa on March 10 values, respectively, with sample size N, and (Fig. 2(b)). Winds were low (< 4 m/s) during O is average of measured data. March 8–10, and were weaker than 2 m/s at most times (Fig. 2(c)). Notably, temperature RESULTS AND DISCUSSION was 13–28oC and relative humidity was 42–86% during this period. Spring episodes (March 8–10, 2005) Fig. 3(a) presents simulated surface wind Figs. 2(a)–(c) present a synoptic surface vectors at 12:00 on March 9, 2005; the solid weather chart for March 9 and hourly lines on the right of the plot represent

324 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b) 1030 1025 1020 1015 1010 Pressure (hPa) (hPa) Pressure 1005 0 122412241224 (c) 5

4 Hsiung-Kong Pingtung Chao-Chou

3

2

1 Wind Speed (m/s) Wind

0 0 122412241224 3/8 3/9 3/10

Fig. 2. (a) Synoptic surface weather chart on March 9, (b) hourly variations of pressure, and (c) hourly variations of wind speed on March 8–10, 2005.

mountain elevations − the highest elevation concentration contours of PM10 reveal that a of approximately 2957 m. In Fig. 3(a), the highly polluted location (> 120 μg/m3) was relatively fast onshore winds from the near the costal lands (Fig. 3(b)); the three northwest blew west inland at reduced dark squares in Fig. 3(b) mark the speeds of 1.1–1.5 m/s. The corresponding monitoring sites of Hsiung-Kong, Pingtung

325 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b)

2520000 100

Pingtung 2510000

80 200 200 2500000 120 100 180

160 Hsiung-Kong UTM-N (m) 2490000 Chao-Chou 140

120 100 100 2480000 80 80 60

40 2470000 20

0

160000 170000 180000 190000 200000 210000

UTM-E (m)

Fig. 3. (a) Simulated surface wind vectors, and (b) simulated concentration contours of PM10 at 12:00 on March 9, 2005.

326 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Measurement Simulation 350 300 Hsiung-Kong IOA = 0.86 ) 3 250 R = 0.76 200 g/m μ

( 150 10 100 50 0 0 122412241224 350 300 Pingtung IOA = 0.75 ) PM 3 250 R = 0.61 200 g/m μ

( 150 10 100 PM 50 0 0 122412241224 350 300 Chao-Chou IOA = 0.76 )

3 250 R = 0.52

g/m 200 μ

( 150 10 100 PM 50 0 0 122412241224

3/8 3/9 3/10

Fig. 4. Comparisons between hourly simulations of surface PM10 concentrations and measured concentrations at the Hsiung-Kong, Pingtung and Chao-Chou sites during March 8–10, 2005. and Chao-Chou. More than 75% of northeast of the Hsiung-Kong site; and the 3 Kaohsiung City was covered in PM10 > 200 PM10 concentration was 115 μg/m at the μg/m3. This highly polluted area extended Chao-Chou site, west of the Hsiung-Kong downwind to the south; the PM10 site. The PM10 concentrations were concentration was 85 μg/m3 at Pingtung site, relatively low (< 80 μg/m3) in the rural and

327 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b) 1030

1025

1020

1015

1010 Pressure (hPa) (hPa) Pressure 1005 0 122412241224 (c) 5 Hsiung-Kong Pingtung Chao-Chou 4

3

2

1 Wind Speed (m/s) Wind 0 0 122412241224 10/12 10/13 10/14

Fig. 5. (a) Synoptic surface weather chart on October 13, (b) hourly variations of pressure, and (c) hourly variations of wind speed on October 12–14, 2005. mountainous regions. Pingtung and Chao-Chou sites during March Fig. 4 compares the 3-day hourly 8–10, 2005. These comparisons suggest that simulations of surface PM10 concentrations surface PM10 concentrations were usually with measured values at Hsiung-Kong, high at midnight and in the early morning,

328 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

and were low in afternoon (around vectors at 24:00 on October 24, 2005. The 16:00–18:00). The highest measured (and relatively fast onshore winds blew from the mean) concentration of PM10 was 215 (92) north to south-west or west to the inland at μg/m3 at Hsiung-Kong, 135 (88) μg/m3 at reduced speeds of 2.3–3.3 m/s, and then Pingtung, and 135 (80) μg/m3 at Chao-Chou. merged with downhill winds near the

These high PM10 events were strongly mountain base. The corresponding correlated with high-pressure systems and concentration contours of PM10 indicated weak winds as a high surface pressure that a highly polluted location (> 200 μg/m3) favors the descent of air parcels aloft and near costal land extended from the southern thus typically increases PM10 concentrations part of Kaohsiung City to the west, near the ground. Although predictions of resembling a banana in shape (Fig. 6(b)). model were in some cases higher or lower The PM10 concentration was approximately than measurements, simulated values were 200 μg/m3 at the Hsiung-Kong and Pingtung generally in agreement with measured sites, and about 106 μg/m3 at the Chao-Chou values, with a correlation coefficient of R = site at midnight. 0.52–0.76, and an index of agreement (IOA) Fig. 7 compares the 3-day hourly

= 0.75–0.86. Notably, the agreement simulations of surface PM10 concentrations between predictions and measurements is with measured values at the three sites regarded as good when IOA exceeds 0.5 during October 12–14, 2005. The PM10 (Hurley et al., 2001 and 2003). concentrations were relatively high at midnight and in the early morning. Autumn episodes (October 12–14, 2005) The highest (and mean) measured values Figs. 5(a) to 5(c) present a synoptic were about 236 (123) μg/m3 at Hsiung-Kong, surface weather chart for October 13 and 246 (105) μg/m3 at Pingtung, and 129 (74) hourly variations in pressure and wind speed μg/m3 at Chao-Chou. Similar to the case in for October 12–14, 2005. A high-pressure spring, high PM10 events in autumn were system and north to northwest winds related to high-pressure systems and weak prevailed in southern Taiwan (Fig. 5(a)). winds. The simulated concentrations agreed Surface pressure was 1008.3–1012.1 hPa reasonably well with measured values, with (Fig. 5(b)). Winds were frequently weak (< R = 0.47–0.56 and IOA = 0.69–0.75. 2 m/s), particularly at midnight and in the early morning (Fig. 5(c)). Relatively strong Winter episodes (December 16–18, 2005) winds of roughly 4 m/s were occasionally The synoptic surface weather chart for observed at 12:00–14:00. Notably, December 17 indicates that a high-pressure temperature was 26–32 oC and relative system with northeasterly winds prevailed humidity was 57–88% during this period over southern Taiwan (Fig. 8(a)). During Fig. 6(a) presents simulated surface wind December 16–18, 2005, surface pressure.

329 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b)

40

2520000 40 60

2510000 Pingtung 100

200 2500000 100 180

Chao-Chou 160 Hsiung-Kong 200 UTM-N (m) 2490000 140

120

100 2480000 40 80

60 60

2470000 40

20

0

160000 170000 180000 190000 200000 210000

UTM-E (m)

Fig. 6. (a) Simulated surface wind vectors, and (b) simulated concentration contours of PM10 at 24:00 on October 14, 2005.

330 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Measurement Simulation 350 300 Hsiung-Kong IOA = 0.75 ) 3 250 R = 0.56 g/m

μ 200 (

10 150 100 50 0 0 122412241224

350

) PM 300 Pingtung IOA = 0.74 3 250 R = 0.56 g/m μ

( 200 10 150 PM 100 50 0 0 122412241224

350 300 Chao-Chou IOA = 0.69 ) 3 250 R = 0.47 g/m

μ 200 ( 10 150 PM 100 50 0 12 24 12 24 12 24 10/12 10/13 10/14

Fig. 7. Comparisons between hourly simulations of surface PM10 concentrations and measured concentrations at the Hsiung-Kong, Pingtung and Chao-Chou sites during October 12–14, 2005. varied at 1019.0–1023.9 hPa (Fig. 8(b)), and this period. winds were frequently weak (< 3 m/s) (Fig. Simulated surface wind vectors at 20:00 8(c)). Notably, temperature was 11–25oC on December 17, 2005 indicates that and relative humidity was 44–80% during relatively strong north winds prevailed over

331 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b) 1030

1025

1020

1015

1010 Pressure (hPa) (hPa) Pressure 1005 0 122412241224 (c) 5 Pingtung Chao-Chou 4 Hsiung-Kong

3

2

1

Wind Speed (m/s) Wind 0 0 122412241224 12/16 12/17 12/18

Fig. 8. (a) Synoptic surface weather chart on December 17, (b) hourly variations of pressure, and (c) hourly variations of wind speed on December 16–18, 2005. the ocean (Fig. 9(a)), while relatively weak located near coasts and mountain base, north to northwest winds prevailed inland covering Kaohsiung City, Pingtung and and merged with downhill winds near the Chao-Chou sites. mountain base. Fig. 9(b) reveals that Fig. 10 compares the 3-day hourly 3 high-polluted regions (PM10 > 200 μg/m ) simulations of surface PM10 concentrations

332 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

(a)

(b)

2520000 100

2510000 Pingtung 20 40

2500000 200 100 200 Chao-Chou 180

UTM-N (m) 160 Hsiung-Kong 2490000 140 140

200 120

100 2480000 140 40 80

60

2470000 40

20

0

160000 170000 180000 190000 200000 210000

UTM-E (m)

Fig. 9. (a) Simulated surface wind vectors, and (b) simulated concentration contours of PM10 at 20:00 on December 17, 2005.

333 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Measurement Simulation 350 300

) Hsiung-Kong IOA = 0.85 3 250 R = 0.76 g/m

μ 200 (

10 150

PM 100 50 0 0 122412241224

350 300 IOA = 0.72

) Pingtung 3 250 R = 0.52 g/m

μ 200 (

10 150

PM 100 50 0 0 122412241224

350 300 Chao-Chou

) IOA = 0.74 3 250 R = 0.56 g/m

μ 200 (

10 150

PM 100 50 0 0 122412241224

12/16 12/17 12/18

Fig. 10. Comparisons between hourly simulations of surface PM10 concentrations and measured concentrations at the Hsiung-Kong, Pingtung and Chao-Chou sites during December 16–18, 2005. with measured concentrations at the three concentration was about 283 (164) μg/m3 at sites during December 16–18, 2005. The the Hsiung-Kong site, 250 (174) μg/m3 at 3 PM10 concentrations were relatively high at the Pingtung site, and 243 (152) μg/m at the midnight. The highest (and mean) measured Chao-Chou site. High PM10 pollution events

334 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

in winter were also associated with neighboring areas of Pingtung or high-pressure systems and weak winds. In Chao-Chou site include Kaohsiung City, additional to even relatively higher pressures Kaohsiung County, Tainan City and Tainan and weaker winds, the dry atmosphere in County. Conversely, neighboring areas winter resulted in PM10 concentrations in the (39.1% at Pingtung; 48.7% at Chao-Chou) winter were higher than those in spring and and area sources (47.4% at Pingtung; 39.4% winter. Simulated concentrations also agreed at Chao-Chou) contributed most to PM10 well with measured values, with R = concentrations, followed by traffic 0.52–0.76 and IOA = 0.72–0.85. emissions (5.8–9.0%) and point sources (4.5–6.1%) at the two sites. Since Pingtung Source contributions at three sites and Chao-Chou are downwind of the The contribution of a particular source Kaohsiung area when north or northeasterly type to PM10 concentrations at one site can winds prevail, PM10 from neighboring be estimated by re-running the TAPM model regions contributed about 44% at the two with emissions for that source at that region sites, significantly higher than that (about removed and comparing the simulated 8%) at the Hsiung-Kong (upwind) site. results with those obtained from the original emission inventory. Table 2 shows some CONCLUSIONS similar and dissimilar features of source contributors among the three sites. Point (or Mesoscale simulations using the TAPM industrial) sources contributed most (49.1%) model were performed on the PM10 episodes to PM10 concentrations at Hsiung-Kong site in southern Taiwan in spring, autumn and

(industrial), followed by area sources winter of 2005. Predictions of hourly PM10 (35.0%), neighboring areas (7.8%), and concentrations agree well with measured traffic emissions (8.1%). Note that the values at three monitoring sites neighboring areas of Hsiung-Kong site (Hsiung-Kong, Pingtung and Chao-Chou). include Kaohsiung County, Pingtung County, The synoptic weather chart indicated that Tainan City and Tainan County. The prevailing winds were northwest (spring),

Table 2. Estimated source contributions at industrial, urban, and rural sites.

Point Neighboring Area Sources (%) Line Sources (%) Monitoring Sources (%) Areas (%) Sites Paved Unpaved Gasoline Diesel Others Road Road Vehicles Vehicles Hsiung-Kong 20.7 0.0 14.3 4.4 3.7 49.1 7.8 Pingtung 21.2 5.2 21.0 4.7 4.3 4.5 39.1 Chao-Chou 18.9 5.1 15.4 2.7 3.1 6.1 48.7

335 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

north (autumn), and northeasterly (winter). for financially supporting this work under Meteorological conditions reveal that Contract No. NSC 94-EPA-Z-110-001 and ambient PM10 typically accumulate and NSC 95-EPA-Z-110-001. trigger episodes on days with high-surface pressure and weak winds. REFERENCES Estimations using TAPM model indicate that industrial (point sources) emissions are Arditsoglou, A. and Samara, C. (2005). the dominant contributors (about 49.1%) to Levels of Total Suspended Particulate

PM10 concentrations at the Hsiung-Kong site Matter and Major Trace Elements in in Kaohsiung City, followed by area sources Kosovo: a Source Identification and (about 35.0%) and transport (about 7.8%) Apportionment Study. Chemosphere 59: from neighboring counties, including 669–678. Kaohsing County, Pintung County, and Chang, M.E. and Cardelino, C. (2000). Tainan area. Because Pingtung City and Application of the Urban Airshed Model Chao-Chou town are downwind of to Forecasting Next-day Peak Ozone Kaohsiung City when north or northeasterly Concentrations in Atlanta, Georgia. J. Air winds prevail, the two sites also experience and Waste Manage. Assoc. 50: high pollution events despite the fact that 2010–2024. these two sites have few industrial pollutant Chen, K.S., Ho, Y.T., Lai, C.H. and Chou sources; neighboring transport of PM10 to Y.M. (2003). Photochemical Modeling the two sites is extremely important, and Analysis of Meteorological accounting for about 39.1% to PM10 Parameters During Ozone Episodes in concentrations at Pingtung and 48.7% at Kaohsiung, Taiwan. Atmos. Environ. 37 : Chao-Chou. Since traffic emissions 1811–1823. contributed little (around 7.8%) to PM10 Chen, K.S., Ho, Y.T., Lai, C.H., Tsai, Y.A. concentrations at the three sites, reducing and Chen, S.J. (2004). Trends in

PM10 emissions from industrial sources in Concentration of Ground-Level Ozone Kaohsiung City should be an effective way and Meteorological Conditions during for improving air quality in Kaohsiung City, High Ozone Episodes in Kao-Ping and in downwind areas, such as Pingtung Airshed, Taiwan, J. Air and Waste County. Manage. Assoc. 54: 36–48. Cheng, S., Chen, D., Li, J., Wang, H. and ACKNOWLEDGMENTS Guo, X. (2007). The Assessment of Emission-source Contributions to Air The authors would like to thank the Quality by Using a Coupled Environmental Protection Administration MM5-ARPS-CMAQ Modeling System: and the National Science Council in Taiwan A Case Study in the Beijing Metropolitan

336 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Region, China. Environ. Modelling and International Clean Air and Environment Software 22: 1601–1616. Conference, New Zealand, Clean Air Cheng, M.D., Gao, N. and Hopke, P.K., Society of Australia and New Zealand. (1996). Source Apportionment Study of Luhar, A.K., Galbally, I.E. and Keywood,

Nitrogen Species Measured in Southern M., (2006). Modelling PM10 California in (1987). J. Environ. Eng. 122: Concentrations and Carrying Capacity 183–190. Associated with Woodheater Emissions in Hurley, P.J., Blockley, A. and Rayner, K. Launceston, Tasmania. Atmos. Environ. (2001). Verification of a Prognostic 40: 5543–5557. Meteorological and Air Pollution Model Lin, J.J. and Lee, L.C., (2004). for Year-long Predictions in the Kwinana Characterization of the Concentration and

Industrial Region of Western Australia. Distribution of Urban Submicron (PM10) Atmos. Environ. 35: 1871–1880 Aerosol Particles. Atmos. Environ. 38: Hurley, P. (2002). The Air Pollution Model 469–475. (TAPM) Version 2. Part 1: Technical Park, I.S., Lee, S.J., Kim, C.H., Yoo, C. and Description. CSIRO Atmospheric Lee, Y.H. (2004) Simulating Urban-scale Research, Private Bag 1, Aspendale, Vic Air Pollutants and their Predicting 3195, Australia. Capabilities Over the Seoul Metropolitan Hurley, P., Manins, P., Lee, S., Boyle, R., Area. J. Air and Waste Manage. Assoc. 54: Ng, Y.L. and Dewundege, P. (2003). 695–710. Year-long, High-resolution, Urban So, K.L. and Wang, T., (2004). C3–C12 Airshed Modelling: Verification of Non-methane Hydrocarbons in TAPM Predictions of Smog and Particles Subtropical Hong Kong: Spatial-temporal in Melbourne, Australia. Atmos. Environ. Variations, Source-receptor Relationships 37: 1899–1910. and Photochemical Reactivity. Sci. of Hurley, P.J., Physick, W.L. and Luhar, A.K., Total Environ. 328: 161–174. (2005). TAPM: A Practical Approach to Stauffer, D.R. and N. Seaman, (1994). Prognostic Meteorological and Air Multiscale four Dimensional Data Pollution Modeling. Environ. Modelling Assimilation. J. App. Meteor. 33: and Software 20: 737–752. 416−434. Hurley, P. (2005). The Air Pollution Model Viana. M., Querol, X., Alastuey, A., Gil, J.I. (TAPM) Version 3. Part 1: Technical and Menendez, M. (2006). Identification Description. CSIRO Atmospheric of PM Sources by Principal Component Research Technical NO. 71. Analysis (PCA) Coupled with Wind Johnson, G.M. (1984). A Simple Model for Direction Data. Chemosphere 65: Predicting the Ozone Concentration of 2411–2418. Ambient Air, Proceedings of the 8th Wang, H. and Shooter, D. (2004). Low

337 Wang and Chen, Aerosol and Air Quality Research, Vol. 8, No. 3, pp. 319-338, 2008

Molecular Weight Dicarboxylic Acids in Applications for Exposure Estimates in

PM10 in a City with Intensive Solid Fuel Cohort Studies. Atmos. Environ. 40: Burning. Chemosphere 56: 725–733. 1053–1063 Willmott, C.J., Ackleson, S.G., Davis, R.E., Zawar-Reza, P., Kingham, S. and Pearce, J. Feddema, J.J., Klink, K.M., Legates, D.R., (2005). Evaluation of a Year-long

O’Donnell, J. and Rowe, C.M. (1985). Dispersion Modelling of PM10 Using the Statistics for the Evaluation and Mesoscale Model TAPM for Christchurch, Comparisons of Model. J. Geophys. Res. New Zealand. Sci. Total Environ. 349: 90: 8995–9005. 249–259. Wilson, J.G. and Zawar-Reza, P. (2006). Intraurban-scale Dispersion Modelling of Received for review, June 16, 2008 Particulate Matter Concentrations: Accepted, July 24, 2008

338