Atmospheric Environment 40 (2006) 3409–3418
Aerosol characteristics from the Taiwan aerosol supersite in the
Asian yellow-dust periods of 2002
Chung-Te Lee
a,, Ming-Tung Chuang
a, Chang-Chuan Chan
b,
Tsun-Jen Cheng
b, Song-Lih Huang
ca
Graduate Institute of Environmental Engineering, National Central University, 300 Jhongda Road, Jhongli, Taoyuan 32054, Taiwan
bInstitute of Occupational Medicine and Industrial Hygiene, National Taiwan University, Taipei, Taiwan cInstitute of Environmental Health, National Yang Ming University, Taipei, Taiwan
Received 23 May 2005; accepted 4 November 2005
Abstract
The occurrence of Asian dust storms, and the subsequent transport of yellow dust (YD) greatly influences the air quality of lee-side countries such as Korea and Japan. The dust is also frequently transported in a southward direction by a strong cold high-pressure system that affects the air quality in Taiwan. This study reports the aerosol properties that were monitored continuously at the Taiwan aerosol supersite during YD events in 2002. Based on the observations of meteorology and aerosols, we divided the time interval of a YD event into a before period, during period, and after period. Among the seven observed YD events, the second event was marked with the maximum hourly PM10level at 502 mg m3,
and with the longest during period for a total of 147 h. The averages of the hourly PM10and PM2.510were much higher in
the during period as compared to those in the before period. It is interesting to note that the time lapse in the during period was well correlated with the maximum level of both PM10and PM2.510. It must be noted that the PM2.5levels were
dramatically increased in the after period, which was due to the accumulation of particles influenced by the anticyclonic outflow. The aerosol size distribution in the third YD event verified that supermicron particles dominated in the during period, and that submicron particles were predominant in the before and after periods. For the chemical properties of the aerosols, time series results indicated that sulfates were mostly contributed by the dust transport, and the others were more related to vehicle exhausts. However, they all accumulated in the period of atmospheric stagnancy.
r2006 Elsevier Ltd. All rights reserved.
Keywords: Aerosol supersite; Atmospheric aerosols; Asian yellow dust; Aerosol characterization
1. Introduction
The source regions of the Asian yellow dust (YD) are distributed broadly over the deserts in
North-west China, Inner Mongolia, and Mongolia. Based
on historical records (Natsagdorj et al., 2003), the
incidences of Asian dust storms occur frequently from winter to spring. The YD drawn by a dust storm is transported mostly by a Siberia high to
China (Zhang et al., 2002), Korea (Kim and Park,
2001), Japan (Ma et al., 2001;Mori et al., 1999), and
even further to North America (Uematsu et al.,
www.elsevier.com/locate/atmosenv
1352-2310/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.11.068
Corresponding author. Tel.: +886 3 4227151ext34657; fax: +886 3 4226551.
1983;Tratt et al., 2001). The YD is also transported
by a strong continental high to Taiwan (Lin, 2001)
and Hong Kong (Fang et al., 1999). Based on the
weather during a YD event, as well as the nature of the dust, the arrival of the YD can be characterized
by a rapid increase in PM10 level, an enhancement
of the wind speed, and a reduction in relative humidity. The YD has been reported to induce an inflammation effect on pulmonary hypertensive rats (Lei et al., 2004) and might thus increase the health risk on vulnerable people. In addition, dust particles
degraded visibility by solar attenuation (Kim et al.,
2001), adsorbed sulfur and nitrogen during their
transport (Kim and Park, 2001; Ma et al., 2001),
and posed a major uncertainty in radiative forcing (Boucher and Anderson, 1996; Koloutsou-Vakakis et al., 1999).
Inspired by the concept of the US aerosol supersite program to continuously monitor aerosol properties, the Taiwan Environmental Protection Administration (TEPA) built its own aerosol super-site to investigate health risk and environmental
impact from aerosols in northern Taiwan (Chan
et al., 2002). The Taiwan aerosol supersite (Super-site) started planning in the year 2000 and was fully operational with loaded instruments in March 2002. Aerosol properties were continuously monitored for
PM2.5(particles with a cut-diameter of less than or
equal to 2.5 mm) and PM10 (particles with a
cut-diameter of less than or equal to 10 mm) mass concentrations, aerosol size distribution ranging
from 0.012 to 10 mm, PM2.5 organic and elemental
carbons, PM2.5sulfate, PM2.5nitrate, aerosol
light-scattering coefficient, aerosol black carbon, and
PM2.5total polycyclic aromatic hydrocarbons. This
study reports some of aerosol properties adopted
from this Supersite for YD events in 2002. The objective in this work is to look into the variations of aerosol properties in different time stages of the YD events.
2. Site location and methods 2.1. Site description
Fig. 1shows the geographical location of Taiwan and the Supersite. The Supersite is located on the ground in a park of Taipei Metropolitan Area. Two major highways contribute vehicle emissions to the Supersite, one running east–west 2–3 km to the north and the other 2 km to the south of the Supersite. The Supersite is expected to monitor ambient aerosols representing typical urban air in Taiwan. The instruments adopted in this study for
data reduction are listed inTable 1. The wind speed
(WS), wind direction (WD), and relative humidity (RH) in this study, however, were adopted from a nearby TEPA air quality monitoring station located 1.7 km southwest of the Supersite, due to the unavailability of the meteorological data at the Supersite at that specific time.
2.2. Data collection methods
The tapered element oscillating microbalance
(TEOM) monitor is a US EPA designated PM10
equivalent method (Designation no. EQPM-1090-079), which has valid measurements from above
5 mg m3 to several g m3 (Rupprecht and
Patash-nick, 2002a; Jaques et al., 2004). The PM2.510
values were taken by subtracting the PM2.5 values
out of the PM10values. They were considered to be
North Korea South Korea Japan Pacific Ocean Philippines Hong Kong China TAIWAN China Taipel Taichung Kaohsiung Taiwan Strait
with the similar precision in PM10 measurements.
All instruments for aerosol speciation were
equipped with a PM10 inlet followed by a sharp
cut cyclone to collect ambient PM2.5. Both aerosol
nitrate and sulfate concentration from R&P 8400N Ambient Particulate Nitrate Monitor and 8400S Ambient Particulate Sulfate Monitor have a base
line stability of 0.4 mg m3 and a measurement
resolution of 0.2 mg m3 (Rupprecht and
Patash-nick, 2001a, b). For aerosol carbon measurements,
Rice (2004)indicated the method detection limit of R&P 5400 Ambient Particulate Carbon Monitor by citing the estimate from the manufacture as
0.10 mg C m3. The concentration of organic carbon
(OC) in this study is the concentration of particulate
carbon (mg C m3) in ambient air without
convert-ing into organic matter. Durconvert-ing the analysis phase of the instrument, the temperature of the collector is raised to 340 1C for a period of 780 s for OC detection and 750 1C for 480 s for total carbon (TC)
detection. The CO2 released at the heating of the
collected particles is measured by a non-disperse
infrared (NDIR) CO2 detector (Rupprecht and
Patashnick, 2002b). The concentration of elemental carbon (EC) was obtained by subtracting the OC values out of the TC values. Aerosol size distribu-tion is measured by the PMS PCASP-X Aerosol
Spectrometer (Particle Measuring Systems, 2001).
This spectrometer is capable of sizing particles over a size range from 0.1 to 10.0 mm in 31 size channels. A He–Ne (632.8 nm) laser is installed into the instrument for particle detection with the minimum detectable size at 0.1 mm.
2.3. Data analysis
To compare the aerosol properties in different time periods of a YD event, we divided the time interval of a YD event into before, during, and after periods. The before period is defined as 24 h prior to
the occurrence of a YD event. The occurrence of a
YD event is initiated by the sudden rise of PM2.510,
the change of WD, and the increase of WS. The during period is then characterized by the
inter-mittent rise of PM2.510 and WS and a relatively
steady WD. The termination of the during period is
justified by the fall of PM2.510and the resumption
of WD and WS to the levels in the before period. The after period is dominated by the atmospheric stagnancy brought about by the anticyclonic out-flow following the during period. In 2002, TEPA officially declared eight YD events affecting
Tai-wan’s air quality. Table 2 shows the dates, the
before, the during, and the after periods of each YD event. Among the eight YD events, the Supersite was able to monitor aerosol properties from the second to the eighth YD events.
3. Results and discussion
3.1. Statistics of aerosol properties and
meteorological parameters monitored in YD events in 2002
Table 3 shows the basic statistics of aerosol properties and meteorological parameters for the three time periods of each YD event in this study. Since the strength of a YD event can be indicated by
the rapid rise in PM2.510, we calculated the
maximum and the average hourly PM2.510 levels
in the during period, they were 443 and 81 mg m3,
respectively. By comparing the PM2.510 levels
between the before and the during periods, the enhancement was found to be around seven times and four times for the maximum and the average values, respectively. This exemplifies the impact of particulate matter during the YD events in 2002.
The PM2.5 level, however, exhibited its highest
value in the after period as a result of the poor ventilation under anticyclonic outflow. In addition,
Table 1
Adopted instruments from Taiwan aerosol supersite and their monitoring capabilities
Adopted instrument Monitoring capability
Rupprecht & Patashnick 1400a TEOM monitor PM10and PM2.5mass concentrations
Rupprecht & Patashnick 5400 ambient particulate carbon monitor
PM2.5total carbon, organic carbon, and elemental carbon
concentrations
Rupprecht & Patashnick 8400N particulate nitrate monitor PM2.5nitrate concentration
Rupprecht & Patashnick 8400S particulate sulfate monitor PM2.5sulfate concentration
the mass fraction of major PM2.5 species (car-bons+sulfate+nitrate) varied from 58% in the before period down to 43% in the during period and went back to 64% in the after period. This indicates a shift from the dominance of the secondary aerosol species to dust related species as the air mass changes from the non-dust period to the dust period.
3.2. PM10and PM2.5levels for the three periods of
YD events in 2002
Fig. 2shows the frequency distribution of PM10 levels for the three time periods of the seven YD
events in 2002. It shows that the PM10 level is the
highest in the during period of the second YD event.
Back trajectory analysis (Draxer, 1999) indicates
that the air masses at 2000 and 1500 m were transported from Inner Mongolia through China’s industrial coastline before coming down to Taiwan (not shown). As the barometric pressure of the second YD event is the greatest among all seven YD
events, we believe that the high PM10 level in the
second event is probably related to the strength of the high-pressure system. At the same time, the
during period PM10level is always higher than that
of the before period. Among the seven YD events observed from the Supersite, the fourth and fifth ones were the least two severe events due to the rains in the during period. A similar rain scavenging effect is also found for the lowest level
of PM10in the after period of the sixth event.Fig. 3
shows the frequency distribution of the PM2.5levels
for the three time periods of the YD events in
2002. It is worth noting that the after period PM2.5
level is always higher than that of the other two periods, except for the sixth event which was due to rains. It should also be noted that the
highest PM2.5 is in the after period of the third
event, while the highest PM10is in the during period
of the second event. This indicates that the intensity of the YD event is not crucial in the
determina-tion of the PM2.5 level in the after period.
Since PM2.5 poses much more of a health concern
than PM10, we must remind people of the health
threat during the after period of a YD event. The weather in the after period has frequently been affected by an anticyclonic outflow. When a high-pressure system moves in from the Asian continent, it is usually a cold high. However, as this continental high moves out from the Asian continent to the West Pacific, it changes from cold to warm. At the same time, the peripheral circula-tion of the high-pressure system turns from south-erly direction to a westsouth-erly direction. This is characterized by a northeastern wind when this anticyclonic outflow reaches the northern part of Taiwan’s coast. Since the weather under an antic-yclonic outflow is stagnant and warm, it is considered to provide a favorable environment for the formation of secondary aerosols. Therefore, it is the poor atmospheric ventilation caused by the subsidence of the air mass, which accounts for the
accumulation of PM2.5.
Table 2
The three time periods for each Asian yellow-dust event affecting Taiwan in 2002 Event Event datesa Periods of the yellow-dust event
Before periodb During periodc After periodd
1 2/11–2/12 2/10 11:00–2/11 11:00 (24)e 2/11 11:00–2/12 11:00 (24) 2/12 19:00–2/13 12:00 (17) 2 3/6–3/9 3/5 01:00–3/6 01:00 (24) 3/6 01:00–3/12 04:00 (147) 3/12 04:00–3/12 20:00 (40) 3 3/18–3/20 3/16 23:00–3/17 23:00 (24) 3/17 23:00–3/19 22:00 (47) 3/19 22:00–3/21 10:00 (36) 4 3/23–3/24 3/22 01:00–3/23 01:00 (24) 3/23 01:00–3/23 20:00 (19) 3/24 03:00–3/25 0:00 (21) 5 3/31–4/1 3/30 06:00–3/31 05:00 (24) 3/31 05:00–4/1 1:00 (20) 4/1 05:00–4/3 09:00 (52) 6 4/8–4/10 4/8 08:00–4/9 08:00 (24) 4/9 08:00–4/11 1:00 (41) 4/11 01:00–4/12 01:00 (24) 7 4/11–4/15 4/11 20:00–4/12 19:00 (24) 4/12 19:00–4/14 03:00 (32) 4/14 03:00–4/16 08:00 (53) 8 4/17–4/19 4/17 04:00–4/18 03:00 (24) 4/18 03:00–4/18 10:00 (7) 4/20 02:00–4/21 08:00 (30) a
The event dates indicate the predicted time period of the yellow-dust event from Taiwan EPA.
b
The before period is defined as 24 hours before the occurrence of a YD event.
cThe during period is justified by the persistence of wind direction (WD) change, high wind speed (WS), and the rise and the fall of
PM2.510(the difference between PM10and PM2.5).
dThe after period is based on the duration of atmospheric stagnancy brought about by an anticyclonic outflow. eIt is noted that the number in parenthesis shows the duration of hours in each time period.
In order to investigate the coarse particle contribution during YD events, we compared the
PM2.510values with the PM2.5values in Fig. 4. It
Table 3
Basic statistics of aerosol properties and meteorological parameters monitored in yellow-dust events in 2002
Measurement item Unit Duration of time (h) Average(7standard deviation) Max. Min. Percentage (%) in PM2.5
Before period PM2.510 a mg m3 157 18(713) 63 2 — PM2.5 mg m3 162 26(715) 75 3 100 TC–PM2.5 mg C m3 126 7(73) 18 2 28 OC–PM2.5 mg C m3 126 6(72) 13 2 21 EC–PM2.5 mg C m3 126 2(71) 6 o1 7 OC/EC–PM2.5 Ratio 126 4(72) 15 2 — Sulfate–PM2.5 mg m3 144 6(74) 14 1 22 Nitrate–PM2.5 mg m3 161 2(72) 12 o1 8 Temperature 1C 161 21(74) 33 15 —
Wind speed m s1 167 1(71) 3 o0.1 —
RHb % 161 77(714) 94 29 — During period PM2.510a mg m3 308 81(765) 443 12 — PM2.5 mg m 3 314 34(714) 83 4 100 TC–PM2.5 mg C m 3 289 7(74) 29 1 19 OC–PM2.5 mg C m 3 289 5(73) 19 o1 14 EC–PM2.5 mg C m 3 289 2(71) 10 o1 5 OC/EC–PM2.5 Ratio 289 3(71) 8 1 — Sulfate–PM2.5 mg m3 290 7(74) 18 1 19 Nitrate–PM2.5 mg m3 312 1(71) 8 o1 4 Temperature 1C 319 20(74) 29 13 — Wind speed m s1 318 2( 71) 6 o0.1 — RHb % 312 56( 717) 99 5 — After period PM2.510a mg m3 259 24(713) 71 1 — PM2.5 mg m3 268 51(729) 144 5 100 TC–PM2.5 mg C m3 263 17(711) 73 2 33 OC–PM2.5 mg C m3 263 12(77) 40 2 24 EC–PM2.5 mg C m3 263 5(74) 33 o1 9 OC/EC–PM2.5 Ratio 263 3(72) 19 1 — Sulfate–PM2.5 mg m3 256 10(75) 23 1 19 Nitrate–PM2.5 mg m 3 270 6(76) 24 o1 13 Temperature 1C 273 23(74) 33 16 —
Wind speed m s1 268 1(71) 4 o0.1 —
RHb % 269 70(716) 96 28 —
aPM
2.510was obtained from the subtraction of PM2.5from PM10. bRH denotes relative humidity.
550 500 450 400 350 300 250 200 150 100 50 0 Concentration ( µ gm -3) 2B 2D 2A 3B 3D 3A 4B 4D 4A 5B 5D 5A 6B 6D 6A 7B 7D 7A 8B 8D 8A
Fig. 2. PM10frequency distribution for each yellow-dust period
in 2002 (2–8: the number of the yellow-dust event, B: before period, D: during period, A: after period. The horizontal lines in each box represent the 95th, 75th, 50th, 25th, and 5th percentiles, respectively, from top to bottom. The max value (m), min value (.), and average value (’) are also shown.
150 100 50 0 2B 2D 2A 3B 3D 3A 4B 4D 4A 5B 5D 5A 6B 6D 6A 7B 7D 7A 8B 8D 8A Concentration ( µ gm -3)
Fig. 3. PM2.5frequency distribution for each yellow-dust period
clearly shows that PM2.510 was dominant in the
PM10in the during period, except for the fifth event
due to rains. As evident in the PM2.510level for the
during period in Fig. 4, the second and the third
events had the highest PM levels observed in Taiwan in 2002. Statistical analysis shows that a
high correlation is found between PM2.510 and
PM10(r2¼0:96), but that it is low between PM2.5
and PM10(r2 ¼0:18) in the during period. This is in
contrast to the high correlation between PM2.5and
PM10in the before and after periods.
3.3. The relationship between the maximum hourly PM level and the hours of the during period
It is necessary for TEPA to know the duration of a YD period in order to call off the warning once it is issued to the general public. As one can find from
Table 2, the during period of a YD event is variable. Among the YD events in 2002, the during period of the second YD event with 147 h was the longest.
The maximum PM10and PM2.510 levels were also
the highest for the second YD event over all. In contrast, the eighth YD event was the one with the
shortest during period, and its maximum PM10and
PM2.510levels were the smallest. If we compare the
maximum hourly PM10and PM2.510levels with the
hours in the during period inFig. 5we find a strong
correlation between them (r2 ¼0:98 for PM
10 and
0.96 for PM2.510). This implies that one can predict
the duration of a YD event simply by referring to its
maximum hourly PM10 level in the during period.
This finding may help authorities to figure out the duration that the YD will influence the local population.
3.4. Continuous aerosol properties and
meteorological parameters in the second YD event Since the duration was the longest and the dust level was the highest in the second YD event, we report the continuous aerosol properties and meteorological parameters for this event. As the YD was transported to Taiwan by a continental high-pressure system in a southerly direction from the dust source region, we checked the synoptic weather map, and found that on the verge of the start of the second YD event, the cold front reached Taiwan at 14:00 (local time) on 5 March 2002 (not shown). It was this continental high-pressure system which brought the YD to Taiwan. On the arrival of the second YD, the WD changed from west to
northeast, and the WS increased from 1 to 2–3 m s1
as shown inFig. 6(a). However, the influence of the
YD started at 02:00 on 6 March 2002, as determined
by the rapid increase of PM2.510 (Fig. 6(b)). This
rapid accumulation of coarse particles was different
from the dominant PM2.5 observed in the before
period. The up and down of the PM2.510 levels
from March 6–12 indicates an intermittent deposit of dust particles in the during period. This implies that the dust was transported in puffs by the air
mass. Except for the peaks of PM2.510and PM2.5in
the early morning of March 6, the other peaks of
PM2.510 and PM2.5 for the rest of this YD event
0 20 40 60 80 100 120 2B 2D 2A 3B 3D 3A 4B 4D 4A 5B 5D 5A 6B 6D 6A 7B 7D 7A 8B 8D 8A Concentration (µg m -3) PM2.5-10 PM2.5
Fig. 4. Average concentration of PM2.510(the difference between PM10and PM2.5) and PM2.5for each yellow-dust period in 2002 (2–8:
the number of the yellow-dust event, B: before period, D: during period, A: after period).
0 100 200 300 400 500 600 2D 3D 4D 5D 6D 7D 8D 0 20 40 60 80 100 120 140 160 PM10 Max (R-square=0.98) PM2.5-10 Max (R-square=0.96) During period hours
Conce n tr ation (µgm -3) Yellow-Dust events Hours
Fig. 5. Aerosol maximum concentration and the hours in the during period for each dust event in 2002.
occurred at different times. This indicates different
source contributions for PM2.510 and PM2.5 for
this event. Fig. 6(c) shows the hourly variations of
both relative humidity (RH) and WS for the second YD event. It is of interest to note that both parameters were complimentary in the course of
this event. When comparing Fig. 6(b) with (c), it is
evident that high PM2.5levels were associated with
high RH and low WS except for the first peak on March 6. This shows that the accumulation of
PM2.5 in the second YD event was aggravated due
to atmospheric stagnancy and high RH. The airmass on the early morning on March 6 originated from Inner Mongolia and then moved along China’s industrial coastline to Taiwan. It is from the first puff of this air mass moving along this path that we observed a simultaneous rise of both
PM2.510 and PM2.5 when it arrived in Taiwan on
March 6. Fig. 6(d) shows that sulfate is the major
constituent of PM2.5, both in the first puff of the
second YD event and in the other peaks of PM2.5.
Because the level of PM2.5 observed on March 12
was so high, we referred to the back trajectory on this date in order to trace back the movements of
this air mass. The 72-h back trajectory (Draxer,
1999) on March 12 originated in the south of Korea
and moved along the Pacific Ocean for most of the time (not shown). Since the air mass passed a clean
ocean area, no anthropogenic sources were expected
to contribute to the observed high PM2.5 level. On
the other end, the trajectory path showed that the air mass was influenced by the anticyclonic outflow.
The PM2.5sulfates for the peaks, other than those in
the early morning on March 6, were thus due to the atmospheric stagnancy that had resulted from the anticyclonic outflow. Other than the time variations
of PM2.5 sulfate, the PM2.5 nitrate did not show a
high concentration in the dusts, but it peaked in the morning on March 10 and 12. This demonstrates
that the PM2.5nitrate was not transported from the
dusts but had accumulated from local activities.
Fig. 6(e)shows the time variations of PM2.5, PM2.5
OC, and PM2.5EC. The PM2.5OC generally follows
the variations of PM2.5except for the time when the
YD hit in the early morning on March 6. Since
PM2.5was mostly contributed from local activities,
it can be concluded that the PM2.5 OC was not a
major constituent in the transported dusts. The
PM2.5EC shows lower values than the PM2.5OC in
the time period of the second dust event, however, it peaks in the morning of each day, reflecting the influence of vehicle emissions during the rush hours.
The PM2.5EC also shows a higher level in the early
morning on March 10, a phenomenon as a result of the atmospheric temperature inversion. From our
investigation on PM2.5 and its chemical properties
0 2 4 6 8 10 3/5 3/6 3/7 3/8 3/9 3/11 3/12 Date Wind speed (ms -1) 0 100 200 300 400 500
Wind direction (degree) Wind speed Wind direction
0 20 40 60 80 100 120 3/5 3/6 3/7 3/8 3/9 3/10 3/11 3/12 Date PM 2.5 (µg m -3) 0 100 200 300 400 500 PM 2.5-10 (µgm -3) PM2.5-10 PM2.5 0 20 40 60 80 100 120 3/5 3/6 3/7 3/8 3/9 3/10 3/11 3/12 Date RH (%) 0 2 4 6 8 Wind speed (m s -1) RH Wind speed 0 20 40 60 80 100 120 3/5 3/6 3/7 3/8 3/9 3/10 3/11 3/12 Date PM 2.5 (µg m -3) 0 5 10 15 20 25 NO 3-, SO 4 (µg m -3) PM2.5 Nitrate Sulfate PM2.5 0 20 40 60 80 100 120 3/5 3/6 3/7 3/8 3/9 3/10 3/11 3/12 Date PM 2.5 (µg m -3) 0 5 10 15 20 OC, EC (µg C m -3) PM2PM.52.5 OC EC 3/10 (a) (b) (c) (d) (e)
Fig. 6. Time variations of wind speed and wind direction (a), PM2.5and PM2.510(b), wind speed and relative humidity (c), PM2.5, PM2.5
sulfate, and PM2.5nitrate (d), PM2.5, PM2.5organic carbon (OC), and PM2.5elemental carbon (EC) (e) monitored at the Taiwan aerosol
supersite for the second yellow-dust event in 2002. In each graph, ‘‘x’’ denotes the time interval for the before period, ‘‘+’’ stands for the time interval for the during period, and ‘‘o’’ exhibits the time interval for the after period.
in the after period of seven YD events in 2002, their peak values usually occurred at 10:00 a.m. and 1:00 p.m. on each day. The former is due to the contribution of vehicle exhausts from the morning rush hour, and the latter can be attributed to low WS and photochemical reaction, because the daily
maximum O3was greater than the monthly average
of the daily maximum in these time periods. 3.5. Aerosol size distribution in a YD event
To understand the variations in aerosol size distribution during a YD event, we extracted aerosol number size distributions measured by the PMS PCASP-X aerosol spectrometer and converted them into aerosol volume size distributions by assuming that the particles are spherical in shape. Owing to incomplete data retrieval in the second YD event, we choose to show the aerosol size
distributions of the third YD event inFig. 7. For a
rough split by size, the aerosol volume can be divided into submicron and supermicron modes in
Fig. 7. The aerosol volume size spectra of the submicron mode in the during period is similar in amount to the before period. However, in the supermicron mode, the aerosol volume clearly exceeds that of the before and after periods. When taking into consideration the aerosol production mechanism, there are three modes in the aerosol volume size distribution throughout the event period. Two peak diameters at 0.4 and 0.7–0.8 mm, respectively, can be identified in the submicron mode. The 0.4 mm size is considered to have grown from condensation nuclei, and the 0.7–0.8 mm particles are thought to be from droplet reaction (McMurry and Wilson, 1982; Morawska et al.,
1998). In the coarse mode, the peak diameter in the
2–3 mm range must be associated with dusts, since a drastic increase in this mode was observed in the during period. To compare the measured aerosol number count with the collected mass, we divided the optical size by the square root of the particle density to obtain particle size that is equivalent to aerodynamic diameter. We then converted the aerosol volume into mass by multiplying the particle density. By assuming an average density of coarse
particles at 2.6 g cm3, we obtained an average
particle mass concentration at 34 mg m3, and a
maximum value at 62 mg m3. The value of 2.6 g m3
for coarse particles was based on the suggestion from Ranz and Wong (1952) in Okada and Kai (2004) for mineral particles and on the density of
quartz at 2.65 g cm3 (Weast and Astle, 1983).
Similarly, we obtained an average particle mass
concentration at 20 mg m3and a maximum value of
39 mg m3in the fine mode for the average particle
density at 2.0 g cm3. The density of fine particles
chosen at 2.0 g m3was adopted from the density of
the important species in fine particles like
ammo-nium sulfate (1.76 g cm3), ammonium nitrate
(1.73 g cm3) (Weast and Astle, 1983), and the
possible mix of small-sized mineral dusts in the fine particles. The sum of the two numbers in the coarse
and fine modes approximates a PM10 average at
54 mg m3, and a maximum value at 101 mg m3. In
contrast, the average PM10 value was 119 mg m3
and the maximum value was 185 mg m3 for the
during period in the third YD event (Fig. 2).
The converted aerosol mass concentration from 0 5 10 15 20 25 0.1 1 10 Dp (µm) dV/dlogDp ( m 3cm -3) (a) 0 10 20 30 40 50 60 70 0.1 1 10 Dp (µm) dV/dlogDp ( m 3cm -3) (b) 0 10 20 30 40 50 0.1 1 10 Dp (µm) (c) dV/dlogDp ( m 3cm -3)
Fig. 7. Aerosol volume size distributions for the before period (a), during period (b), and after period (c) in the third yellow-dust event from 18 to 20 March 2002.
the number count is around half of that from the R&P 1400a TEOM mass monitor. The irregular shape of yellow dusts deviates from the assumption of spherical shape of particles detected by the optical aerosol spectrometer used in this study. An evaluation of particle shape effect on the conversion of particle mass from particle number count is conducted in this study. A dynamic shape factor
(FD) is normally used to relate physical diameter to
the aerodynamic diameter of an aerosol particle (Davies, 1979). The ratio of particle mass collected
by weighing (WW) and particle mass calculated from
counting (WC) is proportional to the square root of
the ratio between FD
3
and the density of the particles (Yang et al., 2004). For a value of FD at 1.41 (Davies, 1979) and the density of dust particles at
2.6 g m3, we find the value of WW=WC is 1.04.
Therefore, the effect of particle shape cannot account for the large difference between the calculated and measured particle mass concentra-tions in this study. In addition, the index of refraction of quartz, ammonium sulfate, and poly-styrene latex spheres (for the calibration of the PMS PCASP-X Aerosol Spectrometer) are 1.54, 1.52, and
1.59, respectively (Weast and Astle, 1983). They are
not significantly different from each other to explain the difference, either. Therefore, we think the deviation might be due to different detection principles between the two instruments, as well as inefficient counting by the laser aerosol spectro-meter when particles are dense in the flow such as the case with dust.
4. Conclusions
We reported here for the first time the compre-hensive Taiwan aerosol supersite data from the YD events in 2002. Seven out of eight dust events were
monitored to show aerosol mass in PM10and PM2.5
as well as aerosol properties. The arrival of the dust can be identified through the rapid increase in
PM2.510accompanied by an increase in WS and a
relatively steady WD. During the course of the
second dust event, the PM2.510level was observed
to vary in puffs, which characterized the pattern of
the dust deposition. The PM2.5level was consistent
with the PM2.510level for the first puff of dust but
from then on the pattern differed among each other for the rest of the duration of the event. It is worth noting that we found a strong correlation between
duration of an event and the maximum PM10 or
PM2.510 level in that event. For the post dust
period, it was noted that the PM2.5 level increased
dramatically due to atmospheric stagnancy brought about by the anticyclonic outflow. Regarding the
PM2.5 chemical properties, sulfates were mostly
contributed from dust transport while the rest were mainly related to vehicle exhausts. However, it must be noted that they were all enhanced by the atmospheric stagnancy in the post dust period. Finally, the aerosol size spectra for the three time periods of a dust event revealed that submicron particles dominated the pre-dust and post-dust periods, while supermicron particles were greatly enhanced during the dust period. A conversion of aerosol volume into mass by assuming a fixed particle density turned out to be around half of the monitored aerosol mass.
Acknowledgements
We are grateful for the support we received from the TEPA for the Taiwan aerosol supersite project. Although the Taiwan aerosol supersite is the official monitoring site of TEPA, the results of this paper are not peer-reviewed by TEPA, and the mentioning of instrument trade names does not constitute the endorsement of TEPA.
References
Boucher, O., Anderson, T.L., 1996. GCM assessment of the sensitivity of direct climate forcing by anthropogenic sulfate aerosols to aerosol size and chemistry. Journal of Geophysical Research 100, 26117–26134.
Chan, C.-C., Her, K.-Z., Huang, G.-H., Lee, C.-T., Wang, P.-Y., Cheng, T.-J., Huang, S.-L., Wang, C.-L., 2002. Health risk assessment of particulate matter. Taiwan EPA Technical Report, EPA-91-FA11-03-91DF02 (in Chinese).
Davies, C.I., 1979. Particle fluid interaction. Journal of Aerosol Science 10, 477–513.
Draxer, R.R., 1999. Hybrid single-particle lagrangian integrated trajectories (HYSPLIT): Version 4.0- User’s Guide. NOAA Technical Memorandum ERL ARL-230, Air Resources Laboratory, Silver Spring, MD, USA.
Fang, M., Zheng, M., Wang, F., Chim, K.S., Kot, S.C., 1999. The long-range transport of aerosols from northern China to Hong Kong—a multi-technique study. Atmospheric Environ-ment 33, 1803–1817.
Jaques, P.A., Ambs, J.L., Grant, W.L., Sioutas, C., 2004. Field evaluation of the differential TEOM monitor for continuous PM2.5mass concentrations. Aerosol Science and Technology
38 (S1), 49–59.
Kim, B.G., Park, S.U., 2001. Transport and evolution of a winter-time Yellow sand observed in Korea. Atmospheric Environment 35, 3191–3201.
Kim, K.W., Kim, Y.J., Oh, S.J., 2001. Visibility impairment during Yellow sand periods in the urban atmosphere of Kwangju, Korea. Atmospheric Environment 35, 5157–5167. Koloutsou-Vakakis, S., Carrico, C.M., Li, Z., Rood, M.J., Ogren,
J.A., 1999. Characterisation of aerosol properties and radiative forcing at an anthropogenically perturbed continental site. Physics and Chemistry of the Earth (C) 24, 541–546. Lei, Y.-C., Chan, C.-C., Wang, P.-Y., Lee, C.-T., Cheng, T.-J., 2004.
Effects of Asian dust event particles on inflammation markers in peripheral blood and bronchoalvelor lavage in pulmonary hypertensive rats. Environmental Research 95, 71–76.
Lin, T.H., 2001. Long-range transport of yellow sand to Taiwan in spring 2000: observed evidence and simulation. Atmo-spheric Environment 35, 5873–5882.
Ma, C.J., Kasahara, M., Tohno, S., Hwang, K.C., 2001. Characterization of the atmospheric aerosols in Kyoto and Seoul using PIXE, EAS and IC. Atmospheric Environment 35, 747–752.
McMurry, P.H., Wilson, J.C., 1982. Growth laws for the formation of secondary ambient aerosols: implications for chemical conversion mechanisms. Atmospheric Environment 16, 121–134.
Morawska, L., Thomas, S., Bofinger, N.D., Wainwright, D., Neale, D., 1998. Comprehensive characterisation of aerosols in a subtropical urban atmosphere: particle size distribution and correlation with gaseous pollutants. Atmospheric Envir-onment 32, 2467–2478.
Mori, I., Iwasaka, Y., Matsunaga, K., Hayashi, M., Nishikawa, M., 1999. Chemical characteristics of free tropospheric aerosols over the Japan Sea coast: aircraft-borne measure-ments. Atmospheric Environment 33, 601–609.
Natsagdorj, L., Jugder, D., Chung, Y.S., 2003. Analysis of dust storms observed in Mongolia during 1937–1999. Atmospheric Environment 37, 1401–1411.
Okada, K., Kai, K., 2004. Atmospheric mineral particles collected at Qira in the Taklamakan Desert, China. Atmo-spheric Environment 38, 6927–6935.
Particle Measuring Systems Co., Inc., 2001. Operating Manual, Passive Cavity Aerosol Spectrometer Probe PMS Model
PCASP-X. Particle Measuring Systems Co., Inc., Boulder, CO, USA.
Ranz, W.E., Wong, J.B., 1952. Impaction of dust and smoke particles on surface and body collectors. Industrial and Engineering Chemistry 44, 1371–1381.
Rice, J., 2004. Comparison of integrated filter and automated carbon aerosol measurements at Research Triangle Park, North Carolina. Aerosol Science and Technology 38 (S2), 23–36.
Rupprecht and Patashnick Co., Inc., 2001a. Operating Manual, Series 8400N Ambient Particulate Nitrate Monitor. February 2001, Revision A.
Rupprecht and Patashnick Co., Inc., 2001b. Operating Manual, Series 8400S Ambient Particulate Sulfate Monitor. June 2001, Revision A.
Rupprecht and Patashnick Co., Inc., 2002a. Operatin Manual, TEOM Series 1400a Ambient Particulate Monitor. March 2002, Revision B.
Rupprecht and Patashnick Co., Inc., 2002b. Operating Manual, Series 5400 Ambient Carbon Particulate Monitor. January 2002, Revision B.
Tratt, D.M., Frouin, R.J., Westphal, D.L., 2001. April 1998 Asian dust event: a southern California perspective. Journal of Geophysical Research 106, 18371–18379.
Uematsu, M., Duce, R.A., Prospero, J.M., Chen, L., Merrill, J.T., McDonald, R.L., 1983. Transport of mineral aerosol from Asia over the North Pacific Ocean. Journal of Geophysical Research 88, 5343–5352.
Weast, R.C., Astle, M.J. (Eds.), 1983. CRC Handbook of Chemistry and Physics. CRC Press, Inc., Boca Raton, Florida, USA.
Yang, H.-H., Cheng, S.-K., Hsieh, L.-T., 2004. Characterization of nitrate particulate dry deposition by vacuum-deposited thin film reaction method. Atmospheric Environment 38, 1785–1793.
Zhang, X.Y., Cao, J.J., Li, L.M., Arimoto, R., Cheng, Y., Huebert, B., Wang, D., 2002. Characterization of atmo-spheric aerosol over Xi An in the south margin of the Loess Plateau, China. Atmospheric Environment 36, 4189–4199.