Seasonal variation, source, and regional

Environ Monit Assess (2010) 167:265–288DOI 10.1007/s10661-009-1048-9

Seasonal variation, source, and regional representativenessof the background aerosol from two remote sitesin western China

Wenjun Qu · Dan Wang · Yaqiang Wang ·Lifang Sheng · Gang Fu

Received: 28 December 2008 / Accepted: 4 June 2009 / Published online: 25 July 2009© Springer Science + Business Media B.V. 2009

Abstract Using observations from two remotesites during July 2004 to March 2005, we showthat at Akdala (AKD, 47◦ 06′ N, 87◦ 58′ E, 562 masl) in northern Xinjiang Province, there werehigh wintertime loadings of organic carbon (OC),elemental carbon (EC), and water-soluble (WS)SO2−

4 , NO2−3 , and NH+

4 , which is similar to thegeneral pattern in most areas of China and EastAsia. However, at Zhuzhang (ZUZ, 28◦ 00′ N,99◦ 43′ E, 3,583 m asl) in northwestern YunnanProvince, the aerosol concentrations and com-positions showed little seasonal variation except

W. Qu (B) · L. Sheng · G. FuKey Laboratory of Physical Oceanography,Ocean-Atmosphere Interaction and ClimateLaboratory, Department of Marine Meteorology,College of Physical and EnvironmentalOceanography, Ocean University of China,Qingdao 266100, People’s Republic of Chinae-mail: [email protected], [email protected]

W. Qu · Y. WangKey Laboratory of Atmospheric Chemistry,Centre for Atmosphere Watch and Services(CAWAS), Chinese Academy of MetrologicalSciences, China Meteorological Administration,Beijing 100081, People’s Republic of China

D. WangState Key Laboratory of Loess and QuaternaryGeology, Institute of Earth Environment,Chinese Academy of Sciences, Xi’an 710075,People’s Republic of China

for a decreasing trend of OC from August toautumn–winter. Additionally, the OC variationsdominated the seasonal variation of PM10 (par-ticles ≤10 μm diameter) level. Chemical char-acteristics combined with transport informationsuggested sea salt origin of ionic Na+, Mg2+, andCl− at ZUZ. At AKD, ionic Ca2+, Mg2+, Na+, andCl− primarily originated from salinized soil. Fur-thermore, the WS Ca2+ contributions (5.4–6%) tothe PM10 mass during autumn, winter, and earlyspring reflected a constant dust component. Theresults of this study indicated that both sites wereregionally representative. However, the represen-tative regions and scales of these background sitesmay vary seasonally as the regional atmospherictransport patterns change. Seasonal variations inthe background aerosol levels from these two ar-eas need to be considered when evaluating theregional climate effects of the aerosols.

Keywords Atmospheric aerosol background ·Regional representativeness · Seasonal variationand source · Zhuzhang (ZUZ) and Akdala(AKD) · Southwestern and northwestern China

Introduction

Atmospheric aerosols vary greatly over time andspace; therefore, accurate depictions of both theaverage aerosol loadings and their variability are

266 Environ Monit Assess (2010) 167:265–288

important for understanding their large-scale im-pacts on global climate systems. Data representa-tive of regional or global aerosol levels are morevaluable in this context. Observations made atcertain remote sites in which the influences oflocal pollutants are minimal and conditions arerepresentative of a large area are expected to pro-vide reasonable representations of a well-mixedatmosphere and are thus potentially suitable forthe analysis of background aerosol conditions(Hidy and Blanchard 2005). However, because itis difficult to ensure that a site is free of pol-lution disturbances, the regional representative-ness of the aerosols collected at such backgroundsites must be considered carefully. For example,Blanchard et al. (1999) investigated the spatialrepresentativeness and scales of transport for pri-

mary particulate and gas-phase precursors of sec-ondary aerosols in San Joaquin Valley, CA, USA.

Although observations have been made fordecades at Chinese background sites, backgroundaerosol data are still limited in China. Past ob-servations primarily targeted long-term variationsof gaseous species (Wang et al. 2002; Zhou et al.2004), while observations of atmospheric aerosolshave mostly been short term and involved traceelements and WS ions (Ma et al. 2003; Wanget al. 2004). Furthermore, to the authors’ knowl-edge, atmospheric background observations werenot made in remote southwestern and northwest-ern China until 2004 (Qu et al. 2008).

Study has been conducted at two China At-mosphere Watch Network (CAWNET) sites,Zhuzhang (ZUZ) and Akdala (AKD, Fig. 1),

50E 60E 70E 80E 90E 100E40E30E20E10E

50E 60E 70E 80E 90E 100E40E30E20E10E

50N

40N

30N

60N

70N

80N

20N

50N

40N

30N

60N

70N

80N

20N

#########################################################

#####

###

##

##

##

##

##

##

###

###

##########################

############

#####################

#################

###

###

##

##

##

##

##

##

##

##

##

##

##

##

##

##

##

##

##

##

###

##############################

###################################

#####

#####

######

###################

##

##

##

##

##

##

##

##

#######################

####

########

####################

####

#################################################################################################

#################

##########

##############################################################################################

#########################################################################################################################

#########################################################################################################################

##############

########

#################################

####

####

##########

### #

# # # ## # # # # # # # # # # # # # # # # # ##########

########

####

############

#######

##################################################

#####

#####

####

######

####

#######

######

#####

#####

#####

#############################################################################################

############################

#[

#[#[

#####################################

####

###

##

##

##

##

##

##

##

##

##

###

####

################################################

####

#########################################################################################################################

#########################################################

#####

###

####

######

##############################################

#########################################################################################################################

#########################################################################################################################

#########################################################################################################################

########################################

########

############

##########

###########

##############

#################

#########

#[

12

3

4

5

6

7

ZUZ

H

I

F

J

A

BC

D

E

G

AKD

NKNS

UMQCD

KM

Proportion of all trajectories (%)Jul–Aug Autumn

Winter March

40 4212 9.7

4.0 1716 50

0 010 6.5

0 2156 24

0 7.16.1 5.6

18 3.00 0

38 100 4.0

12 4.48.6 17

23 1420 15

0 4.40.3 15

14 2513 25

14 4.11.9 0

16 4.413 3.2

0 4.70 0

0 6.30 1.6

2.0 1.96.1 0

20 3037 24

Fig. 1 Sampling sites and 5-day backward trajectory clus-ters for Zhuzhang (ZUZ; cluster 1–7) and Akdala (AKD;cluster A–J). The shadowed area represents continentalChina. NK, NS, UMQ, CD, and KM denote Novokuznetsk,

Novosibirsk, Urumqi, Chengdu, and Kunming, respec-tively. The proportions of every trajectory cluster in allthe trajectories during different seasons are also presented(adapted after Qu et al. 2008)

Environ Monit Assess (2010) 167:265–288 267

to assess regional aerosol backgrounds. We havedescribed the aerosol composition and influencesfrom regional transport previously (Qu et al.2008). In the present paper, we (1) evaluate me-teorological influences on the aerosol concentra-tion and characterize the seasonal variation of theaerosol populations; (2) identify possible sourcesof the aerosol species based on their chemicalcharacteristics, variation, and inter-relationshipsas well as information regarding their transport;(3) assess the spatial representativeness of thetwo sites.

Methods

The atmospheric aerosol composition was mea-sured at two remote CAWNET sites: (1) ZUZ(28◦ 00′ N, 99◦ 43′ E, 3583 m asl, on the southeastmargin of the Tibetan Plateau) and (2) AKD (47◦06′ N, 87◦ 58′ E, 562 m asl, on the north margin ofthe Zhungaer Basin, Fig. 1). Because the samplingand analysis techniques have been described pre-viously (Qu et al. 2008), only a brief descriptionis given below. From July 2004 to March 2005,72-h (normally from 8:00 to 8:00) bulk PM10 (par-ticles ≤10 μm) samples were collected at bothZUZ and AKD. Meanwhile, daytime (normallyfrom 8:00 to 20:00) bulk TSP (total suspendedparticle) samples were also collected but only atAKD. Whatman quartz microfiber filters (47 mmdiameter; QM/A, Whatman, Maidstone, UK; pre-heated at 800◦C for 4 h to remove contaminants)were used for sampling except for the first 2 weekswhen 47-mm-diameter TeflonTM (PTFE) filters(WTP, Whatman) were used. An electronic mi-crobalance with a sensitivity of 1 μg (ME 5-F,Sartorius AG, Goettingen Germany) was used togravimetrically determine the aerosol mass. Theaerosol-laden filters were stored in Petri dishesenclosed in plastic bags at 4◦C after sampling untilchemical analysis was conducted.

The bulk samples were analyzed for organiccarbon (OC)/elemental carbon (EC) using aDRI Model 2001 Thermal/Optical Carbon An-alyzer following the IMPROVE thermal/opticalreflectance procedure (Chow et al. 1993). Thewater-soluble (WS) ions, including SO2−

4 , NO−3 ,

NH+4 , Ca2+, Mg2+, K+, Na+, NO−

2 , Cl−, and F−,

were measured using a Dionex 600 ion chromato-graph equipped with an electrochemical detector(Dionex ED50A). A proton-induced X-ray emis-sion method was used to determine the concentra-tions of selected trace elements, including As, Br,Ca, Cl, Cr, Cu, Fe, K, Mn, Ni, Pb, S, Se, Sr, Ti,V, Zn, and Zr for quartz filters and an additionalfour elements (Al, Mg, P, Si) for TeflonTM filters.All data were corrected for backgrounds based onthe average of the blank filters (Qu et al. 2008).

Because information of transport was helpfulfor identification of possible sources of the aerosolpopulations and evaluation of the regional repre-sentativeness of the observation sites, hierarchicalclustering results of the 5-day air mass backwardtrajectories to ZUZ and AKD (Qu et al. 2008)were also used in this study. The trajectories werecalculated four times each day (00, 06, 12, and18 UTC) using the NOAA HYSPLIT4 trajectorymodel (Draxler and Hess 1998) and then dividedinto distinct transport clusters by a multivariatestatistical analysis method—trajectory clusteringanalysis (Moody and Galloway 1988; Arimotoet al. 1999). National Center for Environmen-tal Prediction FNL (final analysis) meteorologi-cal field data were used in trajectory calculation,while Ward’s method was used to create hierarchi-cal clusters, and the squared Euclidean distanceswere used to determine the similarities betweenthe trajectories. It is worth noting that errorsof 20% of the distance traveled appear to betypical for trajectories computed from analyzedwind fields (Stohl 1998). The aerosol transportpathways to ZUZ and AKD (Fig. 1) are brieflysummarized in Table 1. Detailed information re-garding the transport pathways can be found inQu et al. (2008).

Principal component analysis (PCA) is widelyused to reduce data (Loska and Wiechuya 2003)and extract a small number of latent factors (prin-cipal components, PCs) for assessment of theassociations (relationships) among the observedaerosol species. PCA is often used as an ex-planatory tool to identify the major sources ofair pollutant emissions. The PCA with VARI-MAX normalized rotation, which can maximizethe variances of the factor loadings across vari-ables for each factor, was applied in this study. Assuggested by the Kaiser criterion (Kaiser 1960),


Table 1 Aerosol transport pathways to Zhuzhang (ZUZ) and Akdala (AKD)

Types of transport pathways Trajectory clusters included, source regions, and potential sources

Zhuzhang (ZUZ) The polluted trajectories The northeasterly Cluster 3 (C-3) from the Sichuan Basin withextensive emissions from coal-combustion and mining

The southeasterly C-2 passed over southeastern Yunnan ProvinceClean transport pathways The southwesterly C-1 and C-4 pathways from coastal regions near the

Bay of Bengal (with comparatively low loadings of aerosol species)Remote clean pathways The southwesterly C-5 from the Arabian Sea, the westerly C-6 from

West Asia, and northwesterly C-7 from East Europe (with relativelylow pollutant burdens)

Akdala (AKD) The major pathways for air The northeasterly C–J and northwesterly C–G passing near majorpollutants Russian industrial cities (NK—Novokuznetsk and NS—Novosibirsk)

with steel production, metallurgy, coal mining, and petroleum aspotential sources

The southeasterly C–A passed over the boundary area betweennorthwestern China and Mongolia with extensive coal and metalmining operations

Clean transport pathways The northwesterly C–H, C–I, and C–F pathways (for clean air) fromhigh latitude areas in eastern Europe and Russia

Dust influenced trajectories The westerly C–B, C–C, C–D, and C–E trajectories, influenced bysalinized dust delivered by the prevailing winds from the Gobi, sandylands, and deserts in eastern Kazakhstan and in the area

only those components with eigenvalues >1.0were retained in this study because they can ac-count for a meaningful amount of variance.

Three approaches were applied to identify thelikely sources of the aerosol species: examina-tion of chemical characteristics, variation andinter-correlations among the species combinedwith information regarding their transport, cal-culation of the EFcrust {enrichment factor calcu-lated relative to crustal rock, EFcrust = (Element/Fe)air/(Element/Fe)crust, using the upper conti-nental crust composition compiled by Taylor andMcLennan 1995} of the elements, and PCA.

Results

Characterization of the well-mixed aerosolsat ZUZ and AKD

OC and EC, WS SO2−4 , NH+

4 , Na+, and Mg2+,and elemental S and K generally exhibited syn-chronous variations throughout the course of thestudy at ZUZ (not shown). In addition, good cor-relations were observed between SO2−

4 , NH+4 , and

K+ in ZUZ PM10 samples (Fig. 4a and b), and

these ions were generally anthropogenic origin.At AKD, similar variations were also observedfor OC and EC and WS SO2−

4 and NH+4 ; mean-

while, elemental Ca, Ti, Fe, and K also variedsynchronously during the early period of the studywhen element measurements were available (notshown). EC showed significant correlations withSO2−

4 and K+ in AKD PM10 samples (Fig. 4c),which suggests that there were contributions fromcoal combustion and biomass burning, two pos-sible major sources of anthropogenic emissionsin the region. Furthermore, there was a strongcorrelation between NH+

4 and SO2−4 (Fig. 4d),

which likely resulted from their coexistence as am-monium sulfate/bisulfate particles. However, sig-nificant correlations were also observed betweenNO−

3 , K+, and SO2−4 (apparently from different

origins) at AKD.It is worth noting that the good inter-

correlations between these aerosol species atZUZ and AKD do not necessarily indicate thatthey originated from an identical source. In fact,this may reflect their covariation as a result ofsynoptic processes and mixing during regionaltransport. Because our study sites were far fromresidential areas and other sources of pollution,


the observed results may not have been duemuch to proximate sources. Measurements of theaerosols at these remote sites can thus be regardedas indicators of regional atmospheric backgroundsthat reflect the average conditions of regionaltransported and well-mixed pollutants from vari-ous sources. The arithmetic mean concentrationsof the major aerosol species at ZUZ and AKD areshown in Figs. 2 and 3.

Moreover, the SO2−4 /EC ratio was 12 for most

of the samples (n = 34), larger than that ratio (5.0)for 12 other samples (Fig. 4c). This may reflect theinfluence of coal combustion on most of the sam-pling days. The K+/EC ratio and SO2−

4 /EC ratioshown in Fig. 4c may be regarded as characteristicof different types of regional emissions. Addition-ally, EC also correlated well with elemental S (r =0.63, n = 22, p < 0.01 significance) in the AKDTSP samples, probably due to the contributionfrom coal combustion.

Carbonaceous aerosol at ZUZ and AKD

EC and total carbon (TC) in the PM10 variedsynchronously at low levels, with no pronouncedseasonality at ZUZ, while semi-volatile organiccarbon (SVOC), which is the organic carbon thatevolved from the filter punch in the He-onlyatmosphere at 120◦C during analysis, showed asummer to winter decrease. Because other SVOCsources are limited in such a background region,this distinct seasonality in SVOC levels suggestthe contribution of biogenic source such as theflourishing vegetation during summer and autumn(Qu et al. 2008). Geron et al. (2006) has veri-fied biogenic sources of isoprene from the forestfloor and canopy in this region. OC dominatedthe ZUZ carbonaceous aerosols in PM10 with acontribution to TC that fluctuated around 90%,while only ∼10% of the TC was composed ofEC. In addition, OC variability was large in

Fig. 2 Seasonal variationof organic carbon (OC),elemental carbon (EC),water-soluble (WS) ions,and selected traceelements in PM10at ZUZ. Other ionsincluded K+, Mg2+, F−,Cl−, and NO−

2 ; otherelements included As, Br,Cl, Cr, Cu, Fe, K, Mn, Ni,Pb, Se, Sr, Ti, V, Zn,and Zr

0

1

2

3

4

5

6

7

8

9

August,2004

Autumn,2004

Winter,2004

ZU

Z P

M10

aer

osol

pop

ulat

ion

conc

entr

atio

n (µ

g/m

3 )

Otherelements

Ca

S

Other ions

Ca2+

Na+

NH4+

NO3"

SO42"

EC

OC

August,2004(n =13)

Autumn,2004(n =31)

Winter,2004(n =30)

0.20 0.11 0.15

(±0.12) (±0.067) (±0.086)

0.10 0.13 0.082

(±0.055) (±0.072) (±0.044)

0.38 0.15 0.19

(±0.38) (±0.14) (±0.084)

0.51 0.48 0.56

(±0.31) (±0.26) (±0.32)

0.65 0.48 0.42

(±0.43) (±0.22) (±0.17)

0.58 0.69 0.72

(±0.41) (±0.34) (±0.28)

0.18 0.083 0.20

(±0.22) (±0.074) (±0.13)

0.38 0.51 0.42

(±0.33) (±0.44) (±0.19)

1.8 1.4 1.7

(±1.2) (±0.67) (±0.54)

0.45 0.30 0.36

(±0.25) (±0.16) (±0.17)

3.7 3.2 2.9

(±1.0) (±0.88) (±0.82)

ZUZ PM10 aerosol population average concentration

(Standard devation) (µg/m3)

Other elements

Ca

S

Other ions

Ca2+

Na+

NH4+

-

-

NO3

SO42

EC

OC


Fig. 3 Seasonal variationof organic carbon (OC),elemental carbon (EC),water-soluble (WS) ions,and selected traceelements in PM10at AKD. Other ionsincluded K+, Mg2+, F−,Cl−, and NO−

2 ; otherelements included As, Br,Cl, Cr, Cu, Fe, K, Mn, Ni,Pb, Se, Sr, Ti, V, Zn,and Zr

0

2

4

6

8

10

12

Jul-Aug,2004

Sep & Nov,2004

Winter,2004

Mar,2005

AK

D P

M10

aer

osol

pop

ulat

ion

conc

entr

atio

n (µ

g/m

3 )

Otherions

Na+

Ca2+

NH4+

NO3"

SO42"

EC

OC

Jul-Aug,2004(n =14)

Sep & Nov,2004(n =10)

Winter,2004(n =27)

Mar,2005(n =6)

0.28 0.37 0.35 0.18

(±0.17) (±0.21) (±0.13) (±0.068)

0.25 0.53 0.53 0.46

(±0.23) (±0.19) (±0.12) (±0.071)

0.17 0.45 0.91 0.36

(±0.13) (±0.52) (±1.4) (±0.16)

0.31 0.19 0.87 0.49

(±0.17) (±0.24) (±0.34) (±0.28)

0.23 0.53 0.79 0.53

(±0.23) (±0.71) (±0.26) (±0.26)

1.5 1.9 4.8 2.4

(±0.54) (±1.4) (±1.8) (±1.5)

0.19 0.25 0.45 0.16

(±0.066) (±0.31) (±0.32) (±0.049)

2.5 2.5 3.3 1.8

(±0.83) (±1.3) (±1.8) (±0.13)

AKD PM10 aerosol population

average concentration (±Standard devation) (µg/m3)

Other ions

Na+

Ca2+

NH4+

-

-

NO3

SO42

EC

OC

summer and autumn (2.2–6.2 μg m−3) than inwinter (1.5–4.7 μg m−3), which was likely due tothe more favorable conditions for SVOC emissionand for secondary organic carbon (SOC) trans-formation during summer and autumn. Note herethat SOC was determined based on OC/EC ratio,

that is, SOC = OC − (OC/EC)min × EC (Castroet al. 1999).

PM10 TC at ZUZ was well correlated with hightemperature organic carbon (HTOC), OC, andEC, but not with SVOC (Fig. 5a and b), whichsuggests that HTOC, OC, and EC were constant

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.0

0.2

0.4

0.6

0.8

1.0NH

4

+=0.17SO4

2−-0.11,r=0.81,P<0.0001,n=68

ZU

Z P

M10

NH

4+ , K+ (

µg/m

3 )

ZUZ PM10

SO4

2− (µg/m3)

K+=0.040SO4

2−+0.016,r=0.67,P<0.0001,n=71

a.

-0 .1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.70.0

0.2

0.4

0.6

0.8

1.0

K+ = 0.19 NH4

+ + 0.056

r =0.62, P<0.0001, n=71

ZU

Z P

M10

K+ (

µg/m

3 )

ZUZ PM10

NH4

+ (µg/m3)

b.

0 2 4 6 80.00.51.01.52.02.53.03.54.0

0 2 4 6 80.00.51.01.52.02.53.03.54.0

0 2 4 6 8 100.00.51.01.52.02.53.03.54.0

K+

NH4

+=0.19SO4

2−-0.055,r=0.95,P<0.0001,n=57

AK

D P

M10

NH

4+ , NO

3− , K+ (

µg/m

3 )

NH4

+

NO3

-

AKD PM10

SO4

2− (µg/m3)

NO3

−=0.14SO4

2−+0.14,r=0.63,P<0.0001,n=57

K+=0.017SO4

2-+0.045,r=0.67,P<0.0001,n=57

d.

0.0 0.5 1.0 1.5 2.0 2.50

3

6

9

12

15

0.0 0.5 1.0 1.5 2.0 2.50

3

6

9

12

15

0.0 0.5 1.0 1.5 2.0 2.50

3

6

9

12

15

PM10

SO4

2− = 5.0 EC + 0.61

r =0.94, P<0.0001, n=12

AKD PM10

EC (µg/m3)

PM10

SO4

2− = 12 EC + 0.67

r =0.92, P<0.0001, n=34

AK

D P

M10

SO

42 −, K

+ (

µg/m

3 )

PM10

K+ = 0.14 EC + 0.05

r =0.90, P<0.0001, n=45

c.

Fig. 4 Linear correlation between ionic NH+4 , K+, and

SO2−4 (a) between ionic K+ and NH+

4 (b) in PM10 atZhuzhang (ZUZ); linear correlation between ionic SO2−

4 ,K+, and elemental carbon (EC; c), between ionic NH+

4 ,NO−

3 , K+, and SO2−4 (d) in PM10 at Akdala (AKD). The

good inter-correlations between the aerosol populationsapparently from different sources indicate their synchro-nous variations, suggesting well mixing of these aerosolcomponents during synoptic process and regional transport


1 2 3 4

01234567

1 2 3 4 5 6 7

01234567

EC y = 0.14x - 0.13, r =0.77

OC y = 0.87x - 0.13, r =0.99

ZUZ PM10

Carbon species (µg/m3) (n=70, P<0.0001)

b.

1 2 3 4 5 6 7

01234567

1 2 3 4 5 6 7

01234567 SVOC y = 0.12x - 0.12, r =0.48

ZUZ PM10

Carbon species (µg/m3) (n=70, P<0.0001)

ZUZ PM10

TC (µg/m3) ZUZ PM10

TC (µg/m3)

HTOC y = 0.74x + 0.23, r =0.95a.

Fig. 5 Correlations between semi-volatile organic carbon(SVOC), high temperature organic carbon (HTOC), andtotal carbon (TC) concentrations (a) between organic car-

bon (OC), elemental carbon (EC), and TC concentrations(b) in PM10 at Zhuzhang (ZUZ) during August 2004–February 2005

TC components. Here, HTOC is the carbon frac-tions that evolved from the filter punch in a He-only atmosphere at 250◦C, 450◦C, and 550◦C plusthe pyrolyzed organic carbon during the analysis,which is equal to OC minus SVOC. SVOC wasquite variable, which may have been due to chang-ing meteorological conditions. Indeed, the highlyvariable weather during summer and autumn atZUZ may result in variations in solar radiation,ambient/ground temperature, and soil humidity,which can influence SVOC emission from the veg-etation and soil.

At AKD, PM10 OC and EC varied synchro-nously, with evidently high concentrations occur-ring during winter (similar to WS SO2−

4 ); this waslikely due to increased anthropogenic emissionsfrom domestic heating (Qu et al. 2008). The varia-tions of OC, EC, and SO2−

4 levels also increasedduring winter, as indicated by ranges of 1.8–10.1, 0.12–1.5, and 2.0–8.6 μg m−3, respectively,during winter, and 1.7–3.8, 0.087–0.27, and 0.73–2.8 μg m−3, respectively, during summer. More-over, OC and EC contributed ∼93% and ∼6.9%of the PM10 TC in August versus ∼88% and∼12% during winter. And, the average OC andEC contributions to the TSP TC were ∼92%and ∼7.9% from July to September.

Meteorological influences on the aerosolconcentration

Meteorological conditions can influence theaerosol concentration in different ways. For ex-

ample, the convection intensity can influence thediffusion and dilution of the pollutants, precip-itation can result in wet scavenging of the par-ticles, and the formation of secondary aerosolsas a result of gas/particle conversion of precur-sor gases (photochemical activity) can be influ-enced by solar radiation, ambient temperature,humidity, and the concentrations of oxidants suchas O3.

Statistics of the August-autumn surface wind atZUZ indicated that the 24 h and 14:00 (local time,normally the most intensive convection period ina day) mean wind velocities for EC polluted days(EC concentration > median EC concentration)were 0.92 and 1.58 m/s, respectively. Conversely,these values were 1.16 and 1.90 m/s, respectively,on EC clean days (EC concentration < medianEC concentration). A comparison of the windvelocities for OC polluted and OC clean days wassimilar. That is, high OC and EC levels were foundto be associated with weak wind and convection,which generally enables pollutants to accumulatein the lower boundary layer. Moreover, high OClevels during weak convection periods may alsoresult from more sufficient SOC transformationdue to the prolonged stagnation of the precursorgases in a comparatively stable atmosphere. Astrong southwest wind vector was observed for ECclean days in the surface wind rose (not shown),which likely reflects the contribution of the cleansouthwesterly trajectories (cluster C-1 and C-4)from coastal regions near the Bay of Bengal(Fig. 1, Table 1).


PM10 measurement at ZUZ exhibited higherlevels of OC and EC (3.3 and 0.35 μg m−3) onnon-precipitation days (compared with 2.8 and0.33 μg m−3 on precipitation days), probably dueto the absence of wet scavenging. However, onnon-precipitation days, meteorological conditionsbecame more favorable for more efficient SOCformation from precursor gases and for increasedSVOC emission from natural sources (e.g., vege-tation and soil), which also contributed to higherOC levels then. Moreover, a unique ∼1-weekfluctuation of filterable NO−

3 in PM10 (averaging0.55 μg m−3 on sunny and partly cloudy daysversus 0.36 μg m−3 on rainy and cloudy days) atZUZ was attributed to variation in the photo-chemical transformation of NOχ and formationof NO−

3 in response to meteorological conditionssuch as ambient humidity, temperature, and solarradiation (Qu et al. 2008).

Interestingly, SVOC concentrations in theAKD PM10 samples were in two specific rangeswith modal values of ∼0.2 and ∼0.55 μg m−3

(Fig. 6a and b). The cause of this SVOC fluc-tuation pattern (Fig. 6a) remains unclear. How-ever, there was a significant correlation betweenthe SVOC concentration and daily averaged sur-face temperature (Fig. 6c). These results suggestthat the ambient temperature (varying with theweather shift) had an effect on SVOC emission,which was primarily from natural biogenic sourcesand the evaporation of organic material from soilin remote regions such as AKD.

SOC formation is expected to be more efficientwhen it is sunny and warm and there are high O3

levels (Turpin and Huntzicker 1991; Lim 2001).There was a significant correlation between theSOC concentration in AKD PM10 and the am-bient O3 concentration (observed synchronouslyduring sampling using a 49C UV PhotometricO3 Analyzer, Thermo Electron, USA; Fig. 6d).Although confined by the limited O3 data avail-able, this correlation confirms that a high O3

level is favorable to SOC formation. SOC dom-inated OC variation (Qu et al. 2009), resultingin a similar significant correlation between thePM10 OC concentration and the ambient O3 level(Fig. 6e).

Seasonal variations of the aerosol populationsat ZUZ and AKD

Zhuzhang

At ZUZ, seasonal variations of the individualaerosol populations were not distinct, with theexception of OC, which decreased from August toautumn–winter and dominated the PM10 variation(Fig. 2). This decline in OC can be at least partiallyattributed to reductions in SVOC emission andin SOC transformation that occurred in responseto the changing seasons. In summer and autumn,SVOC might have strong source strength due toseasonality in the growth cycles of vegetation;meanwhile, the formation of SOC was likely moreefficient when solar radiation favored photochem-ical activity intensified (Dusek 2000).

The high summertime PM10 level may haveresulted from the contribution of SVOC andSOC and regional pollutants delivered by the C-3and C-2 trajectories from vicinal sources to ZUZ(Fig. 1, Table 1). However, in autumn and winter,there was an increased frequency of long-rangeC-5, C-6, and C-7 trajectories for clean air, thecontribution from regional pollutants most likelydecreased. The slightly increase of EC, SO2−

4 , andNH+

4 concentrations in winter when comparedto autumn may have resulted from the increasedregional combustion emissions due to the heatingseason.

As a remote background site, the efficient mix-ing upwind, limited local anthropogenic influence,and possible unobvious seasonality in regionalemissions probably led to indistinct seasonal vari-ations of the aerosol populations at ZUZ.

Akdala

Evident seasonality was found for OC, EC, SO2−4 ,

NO2−3 , and NH+

4 in the AKD PM10 samples, withsignificantly higher levels occurring during winter(Fig. 3). Specifically, the wintertime mean ECconcentration was almost twice that of the sum-mer and autumn concentrations, while the winterSO2−

4 and NH+4 concentrations were almost two

to three times those of the summer and autumn.


10 12 14 16 18 20 22 24 26 28 300.0

0.2

0.4

0.6

0.8

1.0

1.2AKD PM

10:

SVOC = 0.067 Temperature - 0.89r = 0.87, P = 0.002, n = 9

AK

D P

M10

SV

OC

(µg

/m3 )

Average ambient temperature (°C)

c.

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00

5

10

15

20

Fre

quen

ce

AKD PM10

SVOC (µg/m3)

0.55

0.2

b.

28 30 32 34 36 38 40 42 44 46

1.0

1.5

2.0

2.5

3.0

3.5

4.0

AKD PM10

SOC:

y = 0.13x - 2.6, r =0.92, P=0.009, n=6

AK

D P

M1O

SO

C (

µg/m

3 )

AKD ambient O3 (ppb) AKD ambient O

3 (ppb)

d.

28 30 32 34 36 38 40 42 44 461.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5 AKD PM10

OC:

y = 0.15x - 3.1, r =0.94, P=0.005, n=6

AK

D P

M1O

OC

(µg

/m3 )

e.

0.0

0.2

0.4

0.6

0.8

1.0

AK

D P

M10

SV

OC

(µg

/m3 )

Sampling date

a.

14 Aug2004

9 Dec2004

8 Jan2005

7 Feb2005

9 Mar2005

Fig. 6 Time series (a) and statistical distribution (b) ofsemi-volatile organic carbon (SVOC) concentrations inPM10 at Akdala (AKD) during August 2004–March 2005and its relationship with available ambient temperatureduring middle August–middle September 2004 (c); the re-

lationship between PM10 secondary organic carbon (SOC)concentration (d), PM10 OC concentration (e), and avail-able ambient O3 concentration in August 2004 at Akdala(AKD)

The higher winter loadings of these pollutantswere likely due to increased emissions from coalcombustion, biomass burning, and other sourcesof domestic heating in northern Xinjiang Provinceand neighboring areas.

WS Ca2+ has been used as an indicator to evalu-ate the contribution of mineral dust (Seinfeld et al.2004; Arimoto et al. 2004). At AKD, althoughthe wintertime mean WS Ca2+ concentration wasalmost two to five times greater than the summer


and autumn concentrations, its fractions in thePM10 mass were in an accordant range (5.4–6%) during the whole sampling course exceptfor summer (2.2%). These results suggest thatthere is a relatively constant dust contributionin autumn, winter, and early spring. Intensivewind during these seasons might lead to enhancedrelease of crust material in this mid–high lati-tude area. However, inconsistent with the gen-erally acknowledged intensive springtime dustemission from East Asian sources, the mean PM10

Ca2+ concentration in March was lower than theconcentrations in autumn and winter (Fig. 3),probably due to limited measurements in thespring.

The WS Na+ and Ca2+ concentrations (0.69and 0.85 μg m−3) were higher than or comparableto the NO−

3 and NH+4 concentrations (0.84 and

0.17 μg m−3) in the AKD TSP samples, whichwas also true for the PM10 at AKD and ZUZ(Figs. 2 and 3). These consistent characteristicsat ZUZ and AKD differ from the conditions atmost urban sites, where sulfate, nitrate, and am-monium are usually major ions in the aerosol (theNO−

3 and NH+4 concentrations were only lower

than the SO2−4 concentration but often higher

than the other ions). Due to the limited local

anthropogenic influence on the sites and possiblelimited regional transport of nitrate and ammo-nium, which are active in photochemical trans-formation and efficient in deposition, crustal ormarine origin materials might become importantaerosol components in these background areas.This would result in the contributions of Na+ andCa2+ to aerosol WS ions being comparable to thecontributions of NO−

3 and NH+4 .

Difference in the concentration of WS ionsbetween PM10 and TSP at AKD

The difference in the concentration of WS ionsbetween PM10 and TSP at AKD provides usefulinformation regarding their source (Fig. 7). Thiscomparison was based on comparable numbers ofboth size samples in August. Most of the ionicNa+, Mg2+, Ca2+, F−, Cl−, NO−

2 , and NO−3 was

present in coarser particles with proportions inthe PM10 comprising less than 60% of those inthe TSP (Fig. 7b). For example, the WS Ca2+in the PM10 accounted for only 15% of that inthe TSP. The distribution of these ions primar-ily occurring in coarser particles was consistentwith their major crustal origin mentioned before.

Fig. 7 Comparison ofthe concentrations ofwater-soluble (WS) ionsin PM10 and in TSP (a);ratio of PM10-to-TSP WSions’ concentrations(percent; b) at Akdala(AKD)

0

0.5

1

1.5

2

SO42– Ca2+ Na+ NO3– NO2– Cl– K+ Mg2+ F–

Concentr

ation (

µg/m

3)

AKD PM10 Aug-04 mean AKD TSP Aug-04 mean

0

25

50

75

100

SO42– Ca2+ Na+ NO3– NO2– Cl– K+ Mg2+ F–

Perc

enta

ge (

%)

AKD PM10/TSP

SO42–

Ca2+

Na+

NO3–

NO2–

Cl–

K+ Mg

2+F

–

AKD PM10 Aug-2004mean concentration

SO42–

Ca2+

Na+

NO3–

NO2–

Cl–

K+ Mg

2+F

–

AKD TSP Aug-2004 mean concentration

AKD: ratio of PM10-to-TSPWS ions’ concentrations(percent)

a.

b.

1.4

1.9

0.1

5 0

.97

0.2

8

0.4

9

0.2

2

0.8

6

0.0

66 0.5

5

0.0

87

0.2

2

0.0

93

0.0

99

0.0

21

0.0

51

0.0

27

0.0

58

72%

15%

58%

26%

12%

39%

94%

41%

46

%


Conversely, the SO2−4 and K+ concentrations in

the PM10 accounted for 72% and 94% of thosein the TSP, indicating large proportions in PM10.These findings may indicate the presence of sec-ondary sulfate from coal combustion and K+ frombiomass burning.

In summary, the secondary aerosols from an-thropogenic sources (including sulfate and potas-sic particles) were primarily present in the PM10,while crust origin cations such as Ca2+, Mg2+,and Na+ were primarily found in the TSP in thisregion. These findings are similar to the size dis-tribution of aerosol ions in Hungary reported byKrivácsy and Molnár (1998).

Similar variation of the aerosols at ZUZand AKD to the PM10 and SO2 loadsin the regional cities

The concentrations of aerosol populations ob-served at ZUZ and AKD were compared with theconcentrations of pollutants in the regional citiesto determine if the results reflected the variationsin the regional pollutants. For this comparison, thedaily PM10 loads in the regional cities were used,including Kunming (KM, 25◦ 01′ N, 102◦ 41′ E)and Chengdu (CD, 30◦ 40′ N, 104◦ 05′ E), whichare located near ZUZ, and Urumqi (UMQ, 43◦ 47′N, 87◦ 36′ E), which is near AKD (Fig. 8).

Fig. 8 Comparison ofthe measured PM10organic carbon (OC)concentration atZhuzhang (ZUZ) withdaily PM10 and SO2 loadsdeduced from the groundair quality index (AQI)in the regional cities ofKunming (KM, ∼500 kmdistant) and Chengdu(CD, ∼570 km distant; a);comparison of measuredPM10 concentration atAkdala (AKD) with dailyPM10 and SO2 loadsdeduced from the groundAQI in the regional cityof Urumqi (UMQ,∼530 km distant; b)

04-29

04-1

0-28

04-1

2-27

0

1

2

3

4

5

6

7

8

9

10

04-29

04-1

0-28

04-1

2-27

0

50

100

150

200

250

300

350

400

450

500 ZUZ PM

10 OC

0.015 Hz Low Pass Filter of ZUZ PM10

OC

ZU

Z P

M10

OC

(µg

/m3 )

Geomean of KM & CD AQI PM10

& SO2

0.015 Hz Low Pass Filter of Geomean of KM & CD AQI PM10

& SO2

Geo

met

ric m

ean

of K

M &

CD

AQ

I PM

10 &

SO

2 (µ

g/m

3 )

a.

29 Aug2004

28 Oct2004

27 Dec2004

25 Feb2005

04-

8-24

04-23

04-22 05

-2-

20

0

100

200

300

400

500

600

700

800

900

04-

8-24

04-23

04-22 05

-2-

20

0

10

20

30

40

50

60

70

80

90

UMQ AQI PM10

& SO2

0.035 Hz Low Pass Filter of UMQ AQI PM10

& SO2

UM

Q A

QI P

M10

& S

O2

(µg/

m3 )

AKD PM10

0.035 Hz Low Pass Filter of AKD PM10

AK

D P

M10

(µg

/m3 )

24 Aug2004

23 Oct

2004

22 Dec2004

20 Feb2005

b.


The daily PM10 loads of the regional citieswere deduced from the ground air quality in-dex (AQI; State Environmental Protection Ad-ministration, available at http://www.sepa.gov.cn/quality/air.php3) for most of the days for whichPM10 was reported as the principal pollutant.Based on the technological rules related to AQI,the following formula was used to derive the PM10

concentration from AQI:

C = [(I − Ilow)

/(Ihigh − Ilow

)]

× (Chigh − Clow

) + Clow, (1)

where C is the concentration of PM10, Ilow andIhigh are AQI grading limited values that are lowerand larger than I (AQI index), respectively, andChigh and Clow denote the PM10 concentrationscorresponding to Ihigh and Ilow, respectively. Thedaily SO2 loads were also deduced in the samemanner as used to deduce the PM10 for severaldays, during which SO2 was reported as the prin-cipal pollutant. The deduced daily PM10 and SO2

loads were subsequently merged into a continuoustime series.

The concentrations of aerosol populations ob-served at AKD were directly compared with thededuced daily PM10 and SO2 loads in UMQ.However, for ZUZ, the major transport path-ways for air pollutants came from Sichuan Basin(C-3) where CD situated and from southeast-ern Yunnan Province (C-2) where KM situated(Fig. 1, Table 1). Therefore, records in KM andCD were used to calculate the daily geometricmean PM10 and SO2 loads, which were subse-quently compared to the results observed at ZUZ.

The measured PM10 OC concentration at ZUZshowed variation trends similar to the geometricmean time series of the deduced daily PM10 andSO2 loads for KM and CD (Fig. 8a). The mea-sured PM10 concentrations at AKD also covariedwith the deduced daily PM10 and SO2 loads forUMQ (Fig. 8b). Linear regressions between datafrom our monitoring sites (ZUZ and AKD) andthe daily PM10 and SO2 loads deduced from theground AQI in the regional cities (∼500 km dis-tant) yielded correlation coefficients (r) rangingfrom 0.39 to 0.68 (p < 0.01 significance). Thesecomparisons revealed that the aerosols observed

at ZUZ and AKD can reflect variation of theregional pollutants.

Comparison of OC/EC concentrations at ZUZand AKD with other background sites

The observed OC concentrations in PM10 at ZUZand AKD (arithmetic mean ± standard deviation,3.1 ± 0.91 and 2.9 ± 1.6 μg m−3) were com-parable with those of other background sites inSouth Korea, America and Hungary (Table 2).In addition, these values also were similar tothe OC background level in America (∼3 μgm−3; Malm et al. 1994) and comparable with OCconcentrations in the boundary layer (median of5.8 μg m−3) and the free troposphere (medianof 3.9 μg m−3) over the North Pacific (Huebertet al. 2004). Similarly, the EC concentrations atZUZ and AKD (0.34 ± 0.18 and 0.35 ± 0.31 μgm−3, respectively) were comparable to the above-mentioned background sites and Mt. Waliguanin China. However, Lim et al. (2003) reportedlower EC concentrations of 0.09 μg m−3 over thebackground Pacific Ocean. Moreover, the OC andEC concentrations at Lin’an regional backgroundsite were much higher than those at ZUZ andAKD. Lin’an is in Zhejiang Province, which is aheavily populated area in southeastern China thatis subject to strong anthropogenic emissions.

It is worth noting that there are high OC andEC levels at urban sites in India, Pakistan, andBengal (Table 2). As important source regionsof air parcels transported to ZUZ, the impact ofcarbonaceous aerosol emission from these neigh-boring countries to southwestern China requiresfurther investigation.

Discussion

Inter-relationships and possible sourcesof the aerosol populations at ZUZ and AKD

Aerosol ions and trace elements at ZUZ

Sea salt Ionic Na+ was generally the secondmost abundant WS ion following SO2−

4 in ZUZPM10 samples during the course of study. Na+ is

http://www.sepa.gov.cn/quality/air.php3

http://www.sepa.gov.cn/quality/air.php3


Tab

le2

Com

pari

son

ofor

gani

cca

rbon

(OC

)an

del

emen

tal

carb

on(E

C)

conc

entr

atio

nsat

Zhu

zhan

g(Z

UZ

)an

dA

kdal

a(A

KD

)w

ith

the

resu

lts

from

othe

rba

ckgr

ound

site

s

Loc

atio

nP

erio

dC

once

ntra

tion

(μg

m−3

)Si

zeSi

tes

prop

erty

Ref

eren

ces

OC

EC

Zhu

zhan

g,Y

unna

n,C

hina

Aug

,200

43.

7±

1.0a

0.45

±0.

25P

M10

Reg

iona

lbac

kgro

und

site

Thi

sst

udy

Aut

umn,

2004

3.2

±0.

880.

30±

0.16

Win

ter,

2004

2.9

±0.

820.

36±

0.17

Aug

,200

4–F

eb,2

005

3.1

±0.

910.

34±

0.18

Akd

ala,

Xin

jiang

,Chi

naA

ug,2

004

2.5

±0.

830.

19±

0.07

PM

10R

egio

nalb

ackg

roun

dsi

teT

his

stud

yA

utum

n,20

042.

5±

1.3

0.25

±0.

31W

inte

r,20

043.

3±

1.8

0.45

±0.

32A

ug,2

004–

Feb

,200

52.

9±

1.6

0.35

±0.

31L

in’a

n,Z

hejia

ng,C

hina

Nov

,199

942

∼44

3.0∼

3.4

PM

2.5

Reg

iona

lbac

kgro

und

site

Xu

etal

.(20

02)

Feb

–Apr

,200

12.

5P

M2.

5W

ang

etal

.(20

04)

Wal

igua

n,Q

ingh

ai,C

hina

Jul,

1994

–Dec

,199

50.

13∼0

.3G

loba

lbac

kgro

und

site

Tan

get

al.(

1999

)K

osan

,Sou

thK

orea

Apr

,199

92.

8±

0.20

0.09

±0.

01P

M2.

5B

ackg

roun

dsi

teK

imet

al.(

2000

)K

angw

ha,S

outh

Kor

eaA

pr,1

999

4.8

±0.

790.

26±

0.06

PM

2.5

Bac

kgro

und

site

Kim

etal

.(20

00)

Loo

kR

ock,

Ten

ness

ee,

2001

1.9∼

4.0

0.55

∼0.6

5P

M2.

5B

ackg

roun

dsi

teT

anne

ret

al.(

2004

)A

mer

ica

Bud

apes

t,H

unga

ryA

pr–M

ay,2

002

2.9∼

5.5

0.19

∼0.6

4P

M2.

0B

ackg

roun

dsi

teSa

lma

etal

.(20

04)

Mum

bai,

Indi

aM

ar,1

999

25.3

12.6

PM

10U

rban

site

Ven

kata

ram

anet

al.(

2002

)L

ahor

e,P

akis

tan

Sep,

1992

–Sep

,199

376

.917

.6T

SPU

rban

site

Smit

het

al.(

1996

)D

haka

,Ban

glad

esh

Apr

–May

,200

145

.722

.0T

SPU

rban

site

Sala

met

al.(

2003

)a A

rith

met

icm

ean

±st

anda

rdde

viat

ion


0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.4

0.8

1.2

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

[Clä /N

a+ m

ole ra

tio] max

=0.6

5

[Clä /N

a+ m

ole

ratio

]=1:

1 line

Mg2+=0.076Na+-0.017,r=0.86,P<0.0001,n=71

ZU

Z P

M10

Mg2+

, Cl– (

µg/m

3 )

ZUZ PM10

Na+ (µg/m3)

Cl-=0.36Na+-0.018,r=0.62,P<0.0001,n=72

a.

Sea salt origin ions:

0.0 0.4 0.8 1.2 1.6

0.00

0.05

0.10

0.15

0.20

0.0 0.4 0.8 1.2 1.6

0.00

0.05

0.10

0.15

0.20 K = 0.28 S + 0.006,

r =0.82, P<0.0001, n=54

ZU

Z P

M10

ele

men

tal K

(µg

/m3 )

ZUZ PM10

elemental S (µg/m3)

K = 0.12 S + 0.0002, r =0.97, P<0.0001, n=18

c.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

0.000

0.005

0.010

0.015

0.020

0.025

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

0.000

0.005

0.010

0.015

0.020

0.025 Pb = 0.043 S + 0.006 r =0.99, P=0.09, n=3

ZUZ PM10

elemental S (µg/m3)

Pb = 0.016 S - 0.0001 r =0.84, P<0.0001, n=54

ZU

Z P

M10

ele

men

tal P

b (µ

g/m

3 )

b.

0.00 0.02 0.04 0.06 0.08 0.10 0.12

0.00

0.04

0.08

0.12

0.16

ZUZ PM10

elemental Fe (µg/m3)

K = 1.2 Fe + 0.004, r =0.77, P<0.0001, n=70

ZU

Z P

M10

ele

men

tal K

(µg

/m3 )

d.

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

0.00

0.01

0.02

0.03

0.04

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

0.00

0.01

0.02

0.03

0.04 Ti = 0.67 Fe - 0.002 r =0.92, P=0.01, n=6

ZUZ PM10

elemental Fe (µg/m3)

Ti = 0.10 Fe - 0.0008 r =0.91, P<0.0001, n=68

ZU

Z P

M10

ele

men

tal T

i (µg

/m3 )

e.

Anthropogenic influenced elements:

Fig. 9 Linear correlations of sea salt origin ions (a) be-tween ionic Cl−, Mg2+, and Na+ in PM10 at Zhuzhang(ZUZ); linear correlations of anthropogenic influenced el-ements between elemental Pb and S (b), between K and S

(c), between anthropogenic influenced K and crust originFe (d), and between Ti and crust origin Fe (e) in PM10 atZUZ

commonly used as an indicator of sea salt(Arimoto et al. 2004; Piel et al. 2006). The close re-lationships between Na+ and Mg2+, Cl− (Fig. 9a)observed in the present study suggested that therewere important marine sources of these ions.These findings were also supported by the highproportion (48.2%) of the trajectories from theArabian Sea (C-5) and from coastal regions nearthe Bay of Bengal (C-1 and C-4; Fig. 1, Table 1).The prevailing southwest monsoon is advanta-geous to sea salt aerosol input by maritime airparcels. The highest concentrations of WS Na+,

Mg2+, and Cl− were observed in samples from theremote southwesterly trajectories from the Ara-bian Sea (C-5; Qu et al. 2008), which is consistentwith their sea salt origin.

The Cl−/Na+ ratio determined from linear re-gression (mass concentration ratio=0.36) deviatesfrom the Cl−/Na+ mole ratio = 1:1 line (Fig. 9a),which suggests that there is a Cl− deficit dueto chemical evolution of the marine sourceaerosols during transport. However, the maxi-mum Cl−/Na+ mole concentration ratio (0.65)could approach 1 under specific meteorology.


Crust origin and anthropogenic influenced ma-terial Six major elements in the ZUZ PM10

samples, including Ca, Fe, Ti, Si, Al, and Mg,exhibited synchronous variations and significantinter-correlations (p < 0.01 significance, r rangedfrom 0.45 to 0.70, n = 74). Their EFcrust valueswere less than 5, indicating that they were typ-ically associated with crustal material. Similarly,elemental Mn and Sr were also primarily of crustorigin.

Conversely, significant correlations (p < 0.05significance, r ranged from 0.29 to 0.96, n = 30–59) were also identified among Zn, Zr, Pb, As,Ni, Se, V, Cu, Cr, and P, but their EFcrust valueswere significantly larger than 5, which suggeststhat they were influenced by non-crustal, anthro-pogenic sources. In addition, the Pb/S ratio formost of the samples (n = 54) was 0.016, but it was0.043 for three other samples (Fig. 9b); the lat-ter samples may reflect contribution of the majorpolluted C-3 trajectories (Fig. 1, Table 1), whichaccounted for 3.9% of the total trajectories andhad the highest Pb concentration (Qu et al. 2008).

Linear regression indicated that the K/S ratiowas 0.28 for most of the ZUZ PM10 samples (n =54), but it was only 0.12 for another group ofsamples (n = 18, Fig. 9c); the higher value of theformer samples was likely due to biomass burning.In addition, there was a higher K/Fe ratio for theZUZ PM10 samples (1.2, Fig. 9d) when comparedwith the same ratio of the crust (0.8, Taylor andMcLennan 1995), which indicates that K was in-fluenced by anthropogenic sources (e.g., biomassburning). The Ti/Fe ratio for most of the samples(n = 68) was 0.10 (Fig. 9e), which was similarto that of the crust (0.09, Taylor and McLennan1995); these results suggest that Ti primarily origi-nated from crustal material. However, six sampleshad a higher Ti/Fe ratio (0.67, Fig. 9e) than thecrust, which indicates that they may have beeninfluenced by industrial emissions such as thosefrom the Sichuan Basin.

Aerosol ions and trace elements at AKD

Soil dust Good correlations were identifiedamong Ca2+, Mg2+, and Na+ in the AKD PM10

samples (Fig. 10a and b). The TSP samples also

showed good correlations between Mg2+ and Na+(Fig. 10c) and significant correlations (p < 0.01significance) between NO−

3 and Ca+ (r = 0.75,n = 22). There were also significant correlationsbetween Cl− and Na+ in both PM10 and TSP(Fig. 10d). Moreover, in the TSP, these ions (typ-ically represented by Mg2+ and Ca+) were alsosignificantly correlated with the presumed crustalorigin elements such as Ca, Fe, Ti, and Mn (rranged from 0.54 to 0.78, n = 22). Taken together,these findings indicate that the aforementionedions may originate from or be influenced by WSsalts in the salinized soil and crustal material.

AKD is located in inland Eurasia; the landcovers around and in the neighborhood (includingXinjiang, most of Mongolia and the neighboringareas in Russia and Kazakhstan) were the aridGobi and desert. Therefore, the WS salts thataccumulated in the surface salinized soil fromsuch land covers were likely entrained into theatmosphere by the strong surface winds, afterwhich they could be transported to our studysite and became important source of aerosol ions.Year-round arid climate and intensive wind maylead to the constant contribution of the WS salts(from the soil) to the aerosol ions. With the ex-ception of three samples collected during dustevents on Nov 30th 2004 and Jan 2nd and Jan5th 2005, consistent relationships among Ca2+,Mg2+, Na+, and SO2−

4 were observed for the PM10

samples (n = 54; Fig. 10a, b, e, and f), whichsupports this contention. Information regardingtransport supplies more evidence. The westerlytrajectories to AKD (C–B, C–C, C–D, and C–E,accounting for 70% of the total trajectories; Fig. 1,Table 1) exhibited comparatively high levels ofCa2+ and Mg2+, as well as quite similar concen-trations of Na+ (Qu et al. 2008). These propertiesmay reflect salinized dust delivered by the prevail-ing winds from sandy lands and deserts in easternKazakhstan and in the area.

Other studies also have reported that aerosolions originated from the salinized soil. Chemicalanalysis and surface structure analysis of the in-dividual particles have shown that the aerosolNa+, Cl−, and SO2−

4 during dust storms in Beijingmost likely originated from the salinized soil thatwas enriched with chloride and sulfate (Zhanget al. 2004). Comparison of dust from the source


0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

16

0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

16

SO4

2–=65Mg2++0.56,r =0.68,P<0.0001,n=54

AKD PM10

Mg2+ (µg/m3)

SO4

2– Dust events samples, n=3

AK

D P

M10

SO

42 – (

µg/m

3 )

e.

0

4

8

12

0 2 4 6 80

4

8

12

SO4

2– = 6.0 Ca2+ + 0.76

r =0.75, P<0.0001, n=54

A

KD

PM

10 S

O42 –

(µg

/m3 )

SO4

2– = 0.42 Ca2+ + 5.0

r =0.99, P=0.03, n=3

AKD PM10

Ca2+ (µg/m3)

f.

Complicated sources of sulfate:

0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

Na+=6.8Mg2++0.17,r =0.80,P<0.0001,n=54

AKD PM10

Mg2+ (µg/m3)

Na+ Dust events samples, n=3

AK

D P

M10

Na+

(µg

/m3 )

a.

0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

0.00 0.05 0.10 0.15 0.20 0.250

4

8

12

AKD PM10

Mg2+ (µg/m3)

AK

D P

M10

Ca2+

(µg

/m3 )

Ca2+ Dust events samples, n=3

Ca2+=7.9Mg2++0.077,r=0.67,P<0.0001, n=54

b.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.00.10.20.30.40.50.60.70.80.91.0

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.60.00.10.20.30.40.50.60.70.80.91.0

AKD PM10

: Cl– = 0.34 Na+ - 0.093,

r =0.66, P<0.0001, n=47 AKD TSP: Cl– = 0.25 Na+ + 0.082,

r =0.78, P<0.0001, n=20

Cl– (

µg/m

3 )

Na+ (µg/m3)

d.

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.180.00.20.40.60.81.01.21.41.61.82.02.2

Na+=8.6Mg2++0.17,r =0.86,P<0.0001,n=21

AK

D T

SP

Na+ , K

+ (µg

/m3 )

AKD TSP Mg2+ (µg/m3)

c.

Salinized soil origin ions:

Fig. 10 Linear correlations of salinized soil origin ionsbetween ionic Na+ and Mg2+ (a), between Ca2+ and Mg2+(b) in PM10, between Na+ and Mg2+ in TSP (c), andbetween Cl− and Na+ in PM10 and TSP (d) at Akdala

(AKD). Linear correlations of sulfate between ionic SO2−4

and Mg2+ (e) and between SO2−4 and Ca2+ (f) in PM10 at

AKD

region and the deposition region has revealed thatthe peak concentrations of Ca2+, SO2−

4 , Cl−, andNa+ in particles larger than 3.3 μm originatedfrom WS salts in the soil dust (Liu et al. 2003).Furthermore, evaluation of aerosols from theTaklimakan Desert has revealed a local soil originof the crustal elements from salts accumulated inthe surface (Makra et al. 2002).

At AKD, elemental Al, Fe, Ca, K, Ti, Mn, andSr in the PM10 and Fe, K, Ca, Ti, Mn, and Se in theTSP were found to be closely related, as indicatedby correlation coefficients of linear regressions

(r) ranging from 0.92 to 0.99 (p < 0.05 signif-icance). The strong correlations between theseelements with Ca and Fe (typical crustal originelements, Zhang et al. 2003) indicated that theymost likely originated from crustal material. El-emental Se also showed strong correlations withthese presumed crustal elements (r ranged from0.96 to 0.98, n = 15, p < 0.01 significance) in TSP.In addition, EFcrust values for these elements (Ca,Fe, K, Ti, Al, Mn, Sr, Si, and Mg) were all closeto unity (maxima less than 5), which is consistentwith their crustal origin.


Complicated source of sulfate It is worth not-ing that the origin of SO2−

4 at AKD was likelycomplicated, including both a crustal source andanthropogenic influence. Good correlations wereobserved among Ca2+, Mg2+, and SO2−

4 in most ofthe AKD PM10 samples (n = 54, Fig. 10e and f).Moreover, ionic SO2−

4 in the TSP was significantlycorrelated with the presumed crustal origin ele-ments such as Ca, Fe, Ti, and Mn (r ranged from0.61 to 0.63, n = 22). These results suggest a crustalsource of SO2−

4 from WS salts in the surfacesalinized soil. Along this same line, the SO2−

4 /WSCa2+ mole concentration ratio for the three PM10

samples collected during dust events (0.18, corre-sponding to a mass concentration ratio of 0.42,Fig. 10f) was comparable to those of the duststorm samples in the Mu Us Desert (0.1, Arimotoet al. 2004). However, the SO2−

4 /WS Ca2+ massconcentration ratio (= 6.0) for most of the samples(n = 54) was approximately 14 times the ratioof samples collected during the three dust events(Fig. 10f), indicating that anthropogenic sourcessuch as coal combustion likely contributed to theionic SO2−

4 in PM10 at AKD for most of thesampling days. The results from a subsequentsource apportion with PCA (“Source apportion ofthe aerosols at ZUZ and AKD”) also support thiscontention, that is, SO2−

4 had a high loading on ananthropogenic component as well as a moderatelyhigh loading on a soil dust component. Ionic NO−

3showed similar consistency, which indicates thatit was substantially loaded on a soil dust compo-nent and moderately loaded on an anthropogeniccomponent.

There are two possible causes of the good inter-correlation between ionic SO2−

4 and NO−3 with

ions of crustal origin such as Ca2+: (1) they orig-inated from the same source, such as the salinizedsoil dust; (2) SO2−

4 was associated with dust par-ticles due to uptaking of SO2 by the dust parti-cles through heterogeneous reactions during theirmixing and transport to the site. These processesinvolved external mixing of various types of pollu-tants and particles, as well as their internal mixing(such as chemical uptake of SO2 and HNO3 bythe dust particles). The internal mixing may resultin coexistence of the aerosol populations fromdifferent sources in the aged aerosol. It is difficultto determine the detailed mechanism of the good

inter-correlation between SO2−4 and ions of crustal

origin (e.g., Ca2+) based our data alone. Never-theless, the larger SO2−

4 /Ca2+ ratio observed formost of our samples compared with the dust stormsamples in the Mu Us Desert tends to supportthe latter inference. Further investigation, such ascomparing the ionic composition of aerosol sam-ples and salinized soil dust samples from aroundthe region, should be conducted to gain a betterunderstanding of this phenomenon.

Anthropogenic influenced material When ele-mental S, Cu, Zn, Ni, and As were evaluated,good correlations were observed between S andOC, NH+

4 , and SO2−4 (r = 0.92, 0.97 and 0.75, n =

8, 14, and 17, respectively, p < 0.01 significance),between S and Cu and Zn (r = 0.93 and 0.86,respectively, n = 17, p < 0.0001 significance), andbetween Zn and Cu and Ni (r = 0.92, n = 17,p < 0.0001 significance) in PM10 at AKD. In ad-dition, the EFcrust values for Zn, Pb, As, Ni, Cu,and Cr (larger than 5) indicated influences fromnon-crustal sources. Taken together, these resultsindicate that these elements may have a closeorigin and represent regional industrial emissions.Trajectories that originated at or passed over theregional industrial and mining areas (C–J, C–G,and C–A) were likely the major pathways forpollutant transport to AKD (Fig. 1, Table 1).There were also significant correlations (p < 0.05significance, r ranged from 0.68 to 0.97, n = 9–22)between elemental Pb, Cu, Zn, As, Ni, and V inthe TSP at AKD, which is consistent with theiranthropogenic origin likely being from regionalindustries.

Source apportion of the aerosols at ZUZand AKD

In a further attempt, PCA with VARIMAX nor-malized rotation was applied to the aerosol chem-ical composition data at the two sites to assistin identification of the possible sources of pollu-tants. Table 3 displays the factor loadings obtainedusing the VARIMAX rotation, as well as theeigenvalues.

PCA identified three components at ZUZ thatexplained 72.4% of the total variance in theaerosol data set. PC1 was highly loaded with SO2−

4


Table 3 Varimax orthogonal rotated principal component matrix for organic carbon (OC), elemental carbon (EC), andwater-soluble ionic species in PM10 aerosol at Zhuzhang (ZUZ) and Akdala (AKD)

Aerosol species Component Communalitiesand parameter 1 2 3 4

Zhuzhang (ZUZ)OC 0.26 0.81 0.14 0.75EC 0.55 0.73 −0.14 0.86OC/EC −0.42 −0.38 0.64 0.73Na+ 0.24 0.34 0.83 0.86NH+

4 0.88 −0.09 −0.19 0.82K+ 0.81 0.11 0.37 0.81Mg2+ 0.49 0.10 0.72 0.77Ca2+ −0.09 0.69 0.38 0.64Cl− −0.04 0.29 0.70 0.58SO2−

4 0.81 0.16 0.15 0.71NO−

3 −0.07 0.65 0.12 0.45Initial eigenvalue 4.10 2.27 1.59Percent of variance 26.9 22.8 22.7Cumulative percent 26.9 49.7 72.4Likely source Fuel combustion Dust/pollutant Sea salt

(coal, biomass, etc.) mixedAkdala (AKD)

OC 0.51 −0.01 0.22 0.83 0.99EC 0.85 0.03 0.39 0.07 0.89OC/EC −0.37 −0.05 −0.19 0.89 0.96Na+ 0.13 0.57 0.71 0.01 0.84NH+

4 0.81 0.46 −0.07 −0.13 0.90K+ 0.80 0.01 0.43 0.05 0.83Mg2+ 0.26 0.86 0.38 −0.06 0.96Ca2+ 0.11 0.93 0.17 −0.01 0.90Cl− 0.12 0.24 0.87 0.03 0.84SO2−

4 0.83 0.51 0.04 −0.07 0.96NO−

3 0.48 0.11 0.64 −0.12 0.67Initial eigenvalue 5.47 1.68 1.45 1.11Percent of variance 31.4 22.4 20.7 13.8Cumulative percent 31.4 53.8 74.5 88.3Likely source Fuel combustion Soil dust Salinized Semi-volatile

(coal, biomass, etc.) soil dust and secondaryorganic carbon

PCA loadings greater than ±0.3 are shown in bold, and those greater than ±0.5 are shown in bold italics

(0.81), NH+4 (0.88), and K+ (0.81), as well as mod-

erately loaded EC (0.55). These results indicatethat PC1 represented an anthropogenic source ofpollutants from fuel combustion (coal, biomass,etc.). PC2 was highly loaded with OC (0.81) andEC (0.73), as well as NO−

3 (0.65) and Ca2+ (0.69),which indicates that it represented a mixture ofanthropogenic pollutants and dust. The OC/ECratios were negatively correlated with PC1 and

PC2 (loadings = −0.42 and −0.38, respectively),which was likely due to the generally low OC/ECratio of the primary carbonaceous aerosols fromanthropogenic emissions. Na+, Mg2+, and Cl−were highly loaded (loadings = 0.83, 0.72, and0.70, respectively) on PC3, which indicates thatit was a sea salt component. This source waslikely associated with maritime air parcels trans-port from the Bay of Bengal and the Arabian Sea


(Fig. 1, Table 1), during which accompanied withformation of secondary OC and diffusion/dilutionof EC; these processes may result in the increasedOC/EC ratio and its positive correlation with PC3(loadings = 0.64).

When the aerosol data set at AKD was consid-ered, PCA revealed four components accountingfor 88.3% of the total variance. PC1 was char-acterized by EC (loading = 0.85), SO2−

4 (0.83),NH+

4 (0.81), and K+ (0.81), while OC loading(0.51) was also moderately high on this compo-nent. These results indicate that PC1 was a typicalanthropogenic source of fuel combustion (coal,biomass, etc.). As discussed above, the OC/ECratio was negatively correlated with this compo-nent. PC2 and PC3 were both ascribed to soildust sources; however, PC2 was characterized byhigh Ca2+ (0.93) and Mg2+ (0.86) loadings as wellas moderately high Na+ (0.57) and SO2−

4 (0.51)loadings, while PC3 was characterized by high Cl−(0.87) and Na+ (0.71) loadings as well as sub-stantially high NO−

3 loading (0.64). These resultsindicate that PC3 may reflect a typical salinizedsoil dust source from the arid Gobi, desert, orsimilar land covers in and around the region. Thelast component (PC4) was only highly loaded withOC (0.83), and the OC/EC ratio was stronglypositively correlated (loading = 0.89) with thiscomponent. These results indicate that PC4 mayreflect the natural SVOC (with no EC emissionaccompanied) and SOC formed during regionaltransport to this remote site.

Regional representativeness of the ZUZand AKD sites

The results of this study indicate that the aerosolscollected at both ZUZ and AKD were region-ally representative, which was supported by fivefactors:

1. The locations of the sites: Both ZUZ and AKDare remote rural sites situated at high alti-tudes; therefore, the aerosols collected werelikely primarily delivered by regional or dis-tant transport with limited impacts from localcirculation and anthropogenic emissions.

2. The well-mixed state of the atmosphere: Co-variation between the aerosol populationsthat appeared to be from different sourcesand constant EC (a typical primary pollutant)contributions to the aerosol mass (5.1% atZUZ versus 3.5% at AKD) implied that theatmosphere sampled was likely well-mixed.

3. The geographical coverage of regional trans-port: Measurements at ZUZ may represent re-gional atmosphere conditions in southwesternChina including Yunnan, Sichuan, Guizhou,Chongqing, and Tibet, in which the trans-port pathways passed over (Fig. 1, Table 1).Measurements at ZUZ may also reflect theinfluences of pollutant input from South Asiancountries such as India, Burma, Bangladesh,and Nepal. The results at AKD can arguablyreflect the atmosphere conditions in north-ern Xinjiang Province and neighboring areasthrough which the transport pathways passed,with also influences of pollutants from upwindcountries.

4. The comparable aerosol levels with otherbackground sites: The observed OC and ECconcentrations at ZUZ and AKD were com-parable to other background sites, indicatingthat baseline aerosol characteristic could beobtained at both sites.

5. The covariation trend of the observed aerosolsin comparison with regional cities: The mea-sured aerosol species at both sites varied syn-chronously with the PM10 and SO2 loads inthe regional cities, indicating that observationat ZUZ and AKD can reflect the variationand regularity of the regional atmosphericcomposition.

Therefore, aerosols collected at ZUZ and AKDcan be expected to be reasonable representationsof the regionally well-mixed atmosphere and apotentially suitable reference for analysis ofregional atmospheric backgrounds.

However, with the change in seasons, the repre-sentative regions (typically the upwind sectors) ofthese background sites varied because of changesin atmospheric transport. Indeed, the source ar-eas, directions, transport distances, and coverageareas of the individual backward trajectories to


ZUZ and AKD differed greatly during differentseasons. For example, the transport pathways toZUZ (Fig. 1, Table 1) during August (late sum-mer) were primarily short, including the south-westerly trajectories from coastal countries nearthe Bay of Bengal (C-1 and C-4, accounted for44% of all trajectories), the southeasterly trajec-tories from over southeastern Yunnan Provinceand south Asian countries (C-2, 38%), and thenortheasterly trajectories from the Sichuan Basinand beyond (C-3, 18%). However, during autumn,winter, and early spring, the pattern of transportdiffered greatly and included more remote trajec-tories from the west (West Asia, C-6), northwest(East Europe, C-7) and southwest (Arabian Sea,C-5); specifically, the proportion of these trajecto-ries increased to 28% and 72% during autumn andwinter, respectively. Other features of wintertimetransport to ZUZ were south-westward migrationand extension of the coverage area of the majorindividual trajectories as well as the absence of thenortheasterly trajectories. At AKD, geographicaldistributions of the trajectories (Fig. 1, Table 1)were generally similar in different seasons. How-ever, there were more trajectories from the north-ern high latitudes during mid–late summer andautumn, more southeasterly trajectories from overthe boundary area between northwestern Chinaand Mongolia during early spring (C–A, 17% ofall trajectories), and southward migration of thecoverage area of the major individual trajectoriesduring winter. The variation in atmospheric trans-port resulted in seasonality of the representativeregions and scales of these sites, which needsfurther assessment.

Further quantitative study is necessary to as-sess the regional representativeness of these twobackground sites. Such studies (similar to thestudy conducted by Blanchard et al. (1999) in SanJoaquin Valley) should include an estimation ofthe scale of transport based on the observationfrom multisites in these regions.

OC/EC ratio used as a means of distinguishingpolluted periods

Because of the inevitable occurrence of distur-bances from natural emissions, anthropogenic ac-tivities, and regional input, identification of the

data with the least perturbation is essential fordetermining aerosol background levels. Hidy andBlanchard (2005) reported that median valuesfor the lower parts of the concentration distrib-utions, as well as the variation and frequency ofthese values derived from long-term and multisiteobservations, provide a better representation ofbackground conditions. We also have suggestedusing the OC/EC ratio as a means of distinguish-ing polluted periods from background conditionscombined with examining the statistic distributionand variation of the data to estimate aerosol back-ground concentrations at ZUZ and AKD (Quet al. 2009). This approach is based on the fact thatthe OC/EC ratio can increase during transport as aresult of secondary transformed SOC being addedto the primary OC fraction, while the OC/EC ratiofor the proximate primary sources was most likelysmaller. The lower OC/EC ratio group of samplesmay reflect influences from proximate sources; forexample, the aerosol populations (including OC,EC, SO2−

4 , NO−3 , and soil dust) from lower OC/EC

ratio samples at both ZUZ and AKD generally ex-hibited higher levels and larger variabilities whencompared with the group of samples with higherOC/EC ratio (Qu et al. 2009). Conversely, thehigher OC/EC ratio group can be expected torepresent a well-mixed atmosphere delivered byregional or distant transport with signatures of theregional background.

Indeed, the concentrations of SO2−4 (a major

aerosol component at both sites, Qu et al. 2008)and the sum of OC + EC + ions exhibited reason-able first-order exponential-decay fitting with theOC/EC ratios at both sites (Fig. 11), similar to thefitting of EC concentrations with OC/EC ratios(Qu et al. 2009). These fitting curves reveal that,as the concentrations of EC, SO2−

4 , and the sumof OC + EC + ions increased, the OC/EC ratiosapproached a constant value that may representthe characteristic OC/EC ratio of the proximateprimary emission sources; while as the OC/ECratios increased, the concentrations of EC, SO2−

4 ,and the sum of OC + EC + ions approached lowvalues, which might reflect their background lev-els. Moreover, SOC/(OC + EC + ions) was foundto be significantly correlated with the OC/EC ratio(Fig. 11), which also indicates that as the SOCproportion increases, the OC/EC ratio increases.


0 10 20 30 40 50 600

1

2

3

4

5

ZUZ PM10

SO4

2-=4.11*exp[-(OC/EC ratio)/7.05]+0.732,

R2=0.527

ZU

Z P

M10

SO

42- (

µg/m

3 )

ZUZ PM10

OC/EC ratio0 5 10 15 20 25 30

0

2

4

6

8

10SO

4

2-=12.3*exp[-(OC/EC ratio)/8.65]-0.179,

R2=0.828 AKD PM10

AK

D P

M10

SO

42- (

µg/m

3 )

AKD PM10

OC/EC ratio

0 10 20 30 40 50 60

4

8

12

16 ZUZ PM10

Y =8.73*exp[-(OC/EC ratio)/7.20]+4.86, R2=0.516

ZU

Z P

M10

(O

C+

EC

+io

ns)

(µg/

m3 )

ZUZ PM10

OC/EC ratio0 10 20 30 40 50 60 70 80 90

0

5

10

15

20

25

30Y =32.8*exp[-(OC/EC ratio)/5.00]+4.05, R2=0.662

AKD PM10

AK

D P

M10

(O

C+

EC

+io

ns)

(µg/

m3 )

AKD PM10

OC/EC ratio

0 10 20 30 40 50 600.2

0.3

0.4

0.5

0.6Y =0.013*(OC/EC ratio)+0.21, r=0.80, P<0.0001, n=59

ZUZ PM10

ZU

Z P

M10

SO

C/(

OC

+E

C+

ions

)

ZUZ PM10

OC/EC ratio

0 10 20 30 40 50 60 70 80 900.0

0.2

0.4

0.6Y =0.017*(OC/EC ratio)+0.041, r=0.84, P<0.0001, n=45

AKD PM10

AK

D P

M10

SO

C/(

OC

+E

C+

ions

)

AKD PM10

OC/EC ratio

a. b.

c. d.

e. f.

Fig. 11 First-order exponential-decay fitting betweenPM10 OC/EC ratio and SO2−

4 concentration at ZUZ (a)and AKD (b) indicates variation trend of sulfate withOC/EC ratio; first-order exponential-decay fitting betweenPM10 OC/EC ratio and Sum (OC + EC + ions) concen-tration at ZUZ (c) and AKD (d) indicates variation trend

of Sum (OC + EC + ions) concentration with OC/ECratio; linear regression fitting between PM10 OC/EC ratioand SOC/Sum (OC + EC + ions) at ZUZ (e) and AKD(f) indicates SOC contribution increases as OC/EC ratioincreases

The relationship between variation in the aerosolpopulations and the OC/EC ratio indicate that itmay be feasible to use the OC/EC ratio to distin-guish samples from regional transport from thosethat have been influenced by proximate emissions,which is helpful for regional atmospheric back-ground assessment. However, further investiga-tion of the general applicability of this approach

for evaluating pollution levels at other remotesites is necessary.

Summary

As part of the evaluation of the two recentlyplanned CAWANET observatory stations (ZUZ


and AKD) in southwestern and northwesternChina, aerosol measurements were made to assessthe regional representativeness of the two sites.Seasonal variations and the sources of the aerosolspecies were also investigated.

1. Our study suggested that the ionic Ca2+,Mg2+, Na+, and Cl− in the PM10 and TSP atAKD had a crustal origin. Microscopic eval-uation of the individual particles may pro-vide morphological evidence in a future study.Additionally, significant correlations betweenSO2−

4 , K+, and EC in the PM10 and betweenEC and elemental S in the TSP revealed con-tributions from coal combustion and biomassburning. The SO2−

4 to WS Ca2+ ratio indicatedthat an anthropogenic source contributed toSO2−

4 on most of the sampling days. More-over, comparison of the PM10 and TSP atAKD implied that crust origin ions like Ca2+,Mg2+, and Na+ were likely enriched in theTSP, whereas the SO2−

4 and K+ in the PM10

accounted for 72% and 94% of those in theTSP, respectively, confirming the influence ofanthropogenic sources.

2. Trace elements in the aerosol were catego-rized into two sources at both sites, elementswith typical crustal origins such as Al, Ca,Fe, Ti, and Si, which showed significant inter-correlations with EFcrust values less than 5, andPb, As, Zn, Ni, V, Cr, and Cu with EFcrust

values greater than 5, which suggested influ-ences from non-crustal, likely anthropogenicsources.

3. At ZUZ, three main source types were de-fined based on PCA (in parentheses are theaerosol species characteristic for each sourcetype). These were (a) anthropogenic emis-sions from fuel combustion such as coal, bio-mass, etc. (SO2−

4 , NH+4 , K+, and EC); (b)

mixtures of anthropogenic pollutants and dust(OC, EC, NO−

3 , and Ca2+); and (c) sea salt(Na+, Mg2+, and Cl−). These components ex-plained 26.9%, 22.8%, and 22.7% of the to-tal variance of the data set, respectively. AtAKD, four main source types were identi-fied: (a) anthropogenic fuel combustion (EC,SO2−

4 , NH+4 , K+, and OC); (b) soil dust (Ca2+,

Mg2+, Na+, and SO2−4 ); (c) salinized soil dust

(Cl−, Na+, and NO−3 ); (d) semi-volatile or-

ganic carbon and secondary organic carbon(high loadings of OC and a high OC/EC ratio).These components explained 31.4%, 22.4%,20.7%, and 13.8% of the total variance,respectively.

4. Based on the locations of the sites, well-mixedstate of the atmosphere, geographical cov-erage of the regional transport, comparableaerosol levels with other background sites,and the probability of the observed aerosolsreflecting the variation of the regional at-mospheric composition, we concluded thatobservations at ZUZ and AKD can reflect thebackground atmosphere. Furthermore, thesesites are also suitable for monitoring the trans-boundary influx of pollutants due to the manytrajectories from abroad.

Finally, it should be noted that PM2.5 measure-ments can better support a study of regionalrepresentativeness than PM10. Unfortunately,PM2.5 observations were not included in thisstudy, but such measurements will be conductedin the future study.

Acknowledgments We are grateful to both anonymousreviewers for their constructive suggestions and com-ments. Wenjun Qu acknowledges members of CAWAS atChinese Academy of Meteorological Sciences for theirsupport and contributions to the project. This work is sup-ported by the National Basic Research Program of China(2006CB403701, 2006CB403702) and the Chinese Ministryof Education’s 111 Project (B07036).

References

Arimoto, R., Snow, J. A., Graustein, W. C., Moody,J. L., Ray, B. J., Duce, R. A., et al. (1999). Influ-ences of atmospheric transport pathways on radionu-clide activities in aerosol particles from over the NorthAtlantic. Journal of Geophysical Research, 104, 21301–21316.

Arimoto, R., Zhang, X. Y., Huebert, B. J., Kang, C. H.,Savoie, D. L., Prospero, J. M., et al. (2004). Chemicalcomposition of atmospheric aerosols from Zhenbeitai,China, and Gosan, South Korea, during ACE-Asia.Journal of Geophysical Research, 109, D19S04.

Blanchard, C. L., Carr, E. L., Collins, J. F., Smith, T. B.,Lehrman, D. E., & Michaels, H. M. (1999). Spatialrepresentativeness and scales of transport during the


1995 integrated monitoring study in California’s SanJoaquin Valley. Atmospheric Environment, 33, 4775–4786.

Castro, L. M., Pio, C. A., Harrison, R. M., & Smith, D. J.(1999). Carbonaceous aerosol in urban and rural Eu-ropean atmospheres: Estimation of secondary organiccarbon concentrations. Atmospheric Environment, 33,2771–2781.

Chow, J. C., Waston, J. G., Pritchett, L. C., Pierson, W.R., Frazier, C. A., & Purcell, R. G. (1993). The DRIthermal/optical reflectance carbon analysis system:Description, evaluation and applications in US airquality studies. Atmospheric Environment, 27A, 1185–1201.

Draxler, R. R., & Hess, G. D. (1998). An overview ofthe HYSPLIT 4 modelling system for trajectories,dispersion, and deposition. Australian MeteorologicalMagazine, 47, 295–308.

Dusek, U. (2000). Secondary organic aerosol—formationmechanisms and source contributions in Europe. In-terim Report IR-00-066. Retrieved from http://www.iiasa.ac.at/.

Geron, C., Owen, S., Guenther, A., Greenberg, J.,Rasmussen, R., Bai, J. H., et al. (2006). Volatile or-ganic compounds from vegetation in southern YunnanProvince, China: Emission rates and some potentialregional implications. Atmospheric Environment, 40,1759–1773.

Hidy, G. M., & Blanchard, C. L. (2005). The midlati-tude North American background aerosol and globalaerosol variation. Journal of the Air & Waste Manage-ment Association, 55, 1585–1599.

Huebert, B., Bertram, T., Kline, J., Howell, S., Eatough, D.,& Blomquist, B. (2004). Measurements of organic andelemental carbon in Asian outflow during ACE-Asiafrom the NSF/NCAR C-130. Journal of GeophysicalResearch, 109, D19S11.

Kaiser, H. F. (1960). The application of electronic comput-ers to factor analysis. Educational and PsychologicalMeasurement, 20, 141–151.

Kim, Y. P., Moon, K. C., Shim, S. G., Lee, J. H., Kim,J. Y., Fung, K., et al. (2000). Carbonaceous species infine particles at the background sites in Korea between1994 and 1999. Atmospheric Environment, 34, 5053–5060.

Krivácsy, Z., & Molnár, Á. (1998). Size distribution ofions in atmospheric aerosols. Atmospheric Research,46, 271–291.

Lim, H. J. (2001). Semi-continuous aerosol carbon mea-surements: Addressing atmospheric progress of localand global concern. Ph.D. thesis, the Graduate School-New Brunswick Rutgers, State University of NewJersey.

Lim, H. J., Turpin, B. J., Russell, L. M., & Bates, T. S.(2003). Organic and elemental carbon measurementsduring ACE-Asia suggest a longer atmospheric life-time for elemental carbon. Environmental Science &Technology, 37(14), 3055–3061.

Liu, M. Z., Wei, W. S., Zhou, H. F., & Yabuki, S. (2003).Physiochemical properties of atmospheric aerosol par-ticles over sand-dust source areas and sedimentary

area in Asia. Journal of Desert Research, 23(4), 408–414 (in Chinese).

Loska, K., & Wiechuya, D. (2003). Application of principlecomponent analysis for the estimation of source ofheavy metal contamination in surface sediments fromthe Rybnik Reservoir. Chemosphere, 51, 723–733.

Ma, J. Z., Tang, J., Li, S. M., & Jacobson, M. Z. (2003). Sizedistributions of ionic aerosols measured at WaliguanObservatory: Implication for nitrate gas-to-particletransfer processes in the free troposphere. Jour-nal of Geophysical Research, 108(D17). doi:10.1029/2002JD003356.

Makra, L., Borbely, K. I., Koltay, E., & Chen, Y. (2002).Enrichment of desert soil elements in Takla Makandust aerosol. Nuclear Instruments and Methods inPhysics Research B, 189, 214–220.

Malm, W. C., Sisler, J. F., Huffman, D., Eldred, R. A.,& Cahill, T. A. (1994). Spatial and seasonal trendsin particle concentration and optical extinction inthe United States. Journal of Geophysical Research,99(D1), 1347–1370.

Moody, J. L., & Galloway, J. N. (1988). Quantifying therelationship between atmospheric transport and thechemical composition of precipitation on Bermuda.Tellus, 40B, 463–479.

Piel, C., Weller, R., Huke, M., & Wagenbach, D. (2006).Atmospheric methane sulfonate and non-sea-salt sul-fate records at the European Project for Ice Cor-ing in Antarctica (EPICA) deep-drilling site in Dron-ning Maud Land, Antarctica. Journal of GeophysicalResearch, 111, D03304. doi:10.1029/2005JD006213.

Qu, W. J., Zhang, X. Y., Arimoto, R., Wang, D., Wang,Y. Q., Yan, L. W., et al. (2008). Chemical compositionof the background aerosol at two sites in southwest-ern and northwestern China: Potential influences ofregional transport. Tellus, 60B, 657–673.

Qu, W. J., Zhang, X. Y., Arimoto, R., Wang, Y. Q., Wang,D., Sheng, L. F., et al. (2009). Aerosol background attwo remote CAWNET sites in western China. Scienceof the Total Environment, 407, 3518–3529.

Salam, A., Bauer, H., Kassin, K., Ullah, S. M., &Puxbaum, H. (2003). Aerosol chemical characteristicsof a mega-city in Southeast Asia (Dhaka-Bangladesh).Atmospheric Environment, 37, 2517–2528.

Salma, I., Chi, X. G., & Maenhaut, W. (2004). Elementaland organic carbon in urban canyon and backgroundenvironments in Budapest, Hungary. AtmosphericEnvironment, 38, 27–36.

Seinfeld, J. H., Carmichael, G. R., Arimoto, R., Conant,W. C., Brechtel, F. J., Bates, T. S., et al. (2004). ACE-ASIA: Regional climatic and atmospheric chemicaleffects of Asian dust and pollution. Bulletin of theAmerican Meteorological Society, 85, 367–380.

Smith, D. J., Harrision, R. M., Luhana, L., Pio, C. A.,Castro, L. M., Tariq, M. N., et al. (1996). Concentra-tions of particulate airborne polycyclic aromatic hy-drocarbons and metals collected in Lahore, Pakistan.Atmospheric Environment, 30, 4031–4040.

Stohl, A. (1998). Computation, accuracy and applica-tions of trajectories—A review and bibliography.Atmospheric Environment, 32, 947–966.

http://www.iiasa.ac.at/

http://www.iiasa.ac.at/

http://dx.doi.org/10.1029/2002JD003356

http://dx.doi.org/10.1029/2002JD003356

http://dx.doi.org/10.1029/2005JD006213


Tang, J., Wen, Y. P., Zhou, L. X., Qi, D. L., Zheng, M.,Trivett, N., et al. (1999). Observational study of blackcarbon in clean air area of western China. Quar-terly Journal of Applied Meteorology, 10(2), 160–170(in Chinese).

Tanner, R. L., Parkhurst, W. J., Valente, M. L., & Phillips,W. D. (2004). Regional composition of PM2.5 aerosolsmeasured at urban, rural and “background” sites inthe Tennessee valley. Atmospheric Environment, 38,3143–3153.

Taylor, S. R., & McLennan, S. M. (1995). The geochemicalevolution of the continental crust. Reviews of Geo-physics, 33, 241–265.

Turpin, B. J., & Huntzicker, J. J. (1991). Secondary for-mation of organic aerosol in the Los Angeles Basin:A descriptive analysis of organic and elemental car-bon concentrations. Atmospheric Environment, 25A,207–215.

Venkataraman, C., Reddy, C. K., Josson, S., & Reddy, M.S. (2002). Aerosol size and chemical characteristicsat Mumbai, India, during the INDOEX-IFP (1999).Atmospheric Environment, 36, 1979–1991.

Wang, G. C., Wen, Y. P., Kong, Q. X., Ren, L. X., &Wang, M. L. (2002). CO2 background concentrationin the atmosphere over the Chinese mainland. ChineseScience Bulletin, 47(14), 1217–1220.

Wang, T., Wong, C. H., Cheung, T. F., Blake, D. R.,Arimoto, R., Baumann, K., et al. (2004). Relationshipsof trace gases and aerosols and the emission charac-teristics at Lin’an, a rural site in eastern China, dur-ing spring 2001. Journal of Geophysical Research, 109,D19S05.

Xu, J., Bergin, M. H., Yu, X., Liu, G., Zhao, J., Carrico,C. M., et al. (2002). Measurement of aerosol chemical,physical and radiative properties in the Yangtze deltaregion of China. Atmospheric Environment, 36, 161–173.

Zhang, X. Y., Gong, S. L., Arimoto, R., Shen, Z. X., Mei,F. M., Wang, D., et al. (2003). Characterization andtemporal variation of Asian dust aerosol from a site inthe northern Chinese deserts. Journal of AtmosphericChemistry, 44, 241–257.

Zhang, X. Y., Zhuang, G. S., & Yuan, H. (2004). Thedried salt-lakes saline soils sources of the dust storm inBeijing: The individual particles analysis and XPS sur-face structure analysis. China Environmental Science,24(5), 533–537 (in Chinese).

Zhou, L. X., Worthy, D. E. J., Lang, P. M., Ernst, M. K.,Zhang, X. C., Wen, Y. P., et al. (2004). Ten years ofatmospheric methane observations at a high elevationsite in western China. Atmospheric Environment, 38,7041–7054.

Documents

Seasonal variation, source, and regional